mirror/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2025-01-17 20:42:30 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions
mariadb/sql/slave.cc

4517 lines
150 KiB
C++
Raw Normal View History

Added timeout for wait_for_master_pos Fixed comparision of log-binary name to handle comparison when file name extension wraps from .999 to .1000 Don't replicate CREATE/DROP DATABASE if wild_xxx_table=database.% is used.
2003-01-25 14:07:51 +01:00
/* Copyright (C) 2000-2003 MySQL AB
open binlog index file *before* tc_log->open() and binlog itself *after*
2005-02-01 15:36:48 +01:00
Import changeset
2000-07-31 21:29:14 +02:00
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
Many files: Changed header to GPL version 2 only
2006-12-23 20:17:15 +01:00
the Free Software Foundation; version 2 of the License.
open binlog index file *before* tc_log->open() and binlog itself *after*
2005-02-01 15:36:48 +01:00
Import changeset
2000-07-31 21:29:14 +02:00
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
open binlog index file *before* tc_log->open() and binlog itself *after*
2005-02-01 15:36:48 +01:00
Import changeset
2000-07-31 21:29:14 +02:00
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
/**
@addtogroup Replication
@{
@file
@brief Code to run the io thread and the sql thread on the
replication slave.
*/
Import changeset
2000-07-31 21:29:14 +02:00
#include "mysql_priv.h"
SCRUM embedded library trimming
2003-01-15 09:11:44 +01:00
Import changeset
2000-07-31 21:29:14 +02:00
#include <mysql.h>
Fixed delete in tables with hidden primary key Remove not used BDB logs on shutdown Don't give warnings for repair on slaves Fixed transaction log files
2000-12-12 03:34:56 +01:00
#include <myisam.h>
laptop commit, syncing with the repostitory. Fixed some bad bugs in replication
2000-11-11 22:50:39 +01:00
#include "slave.h"
BUG#25688 (RBR: circular replication may cause STMT_END_F flags to be skipped): By moving statement end actions from Rows_log_event::do_apply_event() to Rows_log_event::do_update_pos() they will always be executed, even if Rows_log_event::do_apply_event() is skipped because the event originated at the same server. This because Rows_log_event::do_update_pos() is always executed (unless Rows_log_event::do_apply_event() failed with an error, in which case the slave stops with an error anyway). Adding test case. Fixing logic to detect if inside a group. If a rotate event occured when an initial prefix of events for a statement, but for which the table did contain a key, last_event_start_time is set to zero, causing rotate to end the group but without unlocking any tables. This left a lock hanging around, which subsequently triggered an assertion when a second attempt was made to lock the same sequence of tables. In order to solve the above problem, a new flag was added to the relay log info structure that is used to indicate that the replication thread is currently executing a statement. Using this flag, the replication thread is in a group if it is either in a statement or inside a trans- action. The patch also eliminates some gratuitous header file inclusions that were not needed (and caused compile errors) and replaced them with forward definitions.
2007-04-12 08:58:04 +02:00
#include "rpl_mi.h"
#include "rpl_rli.h"
updates for LOAD DATA FROM MASTER + some cleanup of replication code
2001-05-29 03:18:23 +02:00
#include "sql_repl.h"
WL#2472: Renamed Table_filter to Rpl_filter Eliminated very annoying memory leak
2005-03-21 22:09:42 +01:00
#include "rpl_filter.h"
fixes for mysql-test to run without manager a bit more work on failsafe replication
2001-10-11 21:54:06 +02:00
#include "repl_failsafe.h"
Import changeset
2000-07-31 21:29:14 +02:00
#include <thr_alarm.h>
Fix after merge Updated WEEK() and DATE information in the manual.
2002-08-08 15:41:04 +02:00
#include <my_dir.h>
SCRUM: Here is another pack of changes about gathering common client code in sql-common/client.c. Now i symlink the client.c from sql/ and libmysql/. These directories have client_settings.h files to be included to client.c. It contains defines and declarations to compile client.c in appropriate manner. Also i've added include/sql_common.h, containing declarations of what is exported from client.c I removed as many #ifdef-s from client.c as i dared to. I think it's better push it with some extra #ifdef-s now (of course, if everythihg besides it is ok) so other people can check the code.
2003-05-31 12:15:46 +02:00
#include <sql_common.h>
Fix for bug #23775 "Replicated event larger that max_allowed_packet infinitely re-transmits". Problem: to handle a situation when the size of event on the master is greater than max_allowed_packet on slave, we checked for the wrong constant (ER_NET_PACKET_TOO_LARGE instead of CR_NET_PACKET_TOO_LARGE). Solution: test for the client "packet too large" error code instead of the server one in slave I/O thread.
2007-03-16 15:25:20 +01:00
#include <errmsg.h>
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
#include <mysqld_error.h>
Fix for bug #24192 "MySQL replication does not exit server when running out of memory" In case of out-of-memory error received from the master, print the corresponding message to the error log and stop slave I/O thread to avoid reconnecting with a wrong binary log position.
2007-07-11 16:38:45 +02:00
#include <mysys_err.h>
SCRUM embedded library trimming
2003-01-15 09:11:44 +01:00
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
#ifdef HAVE_REPLICATION
#include "rpl_tblmap.h"
BUG#22864 (Rollback following CREATE... SELECT discards 'CREATE TABLE' from log): When row-based logging is used, the CREATE-SELECT is written as two parts: as a CREATE TABLE statement and as the rows for the table. For both transactional and non-transactional tables, the CREATE TABLE statement was written to the transaction cache, as were the rows, and on statement end, the entire transaction cache was written to the binary log if the table was non-transactional. For transactional tables, the events were kept in the transaction cache until end of transaction (or statement that were not part of a transaction). For the case when AUTOCOMMIT=0 and we are creating a transactional table using a create select, we would then keep the CREATE TABLE statement and the rows for the CREATE-SELECT, while executing the following statements. On a rollback, the transaction cache would then be cleared, which would also remove the CREATE TABLE statement. Hence no table would be created on the slave, while there is an empty table on the master. This relates to BUG#22865 where the table being created exists on the master, but not on the slave during insertion of rows into the newly created table. This occurs since the CREATE TABLE statement were still in the transaction cache until the statement finished executing, and possibly longer if the table was transactional. This patch changes the behaviour of the CREATE-SELECT statement by adding an implicit commit at the end of the statement when creating non-temporary tables. Hence, non-temporary tables will be written to the binary log on completion, and in the even of AUTOCOMMIT=0, a new transaction will be started. Temporary tables do not commit an ongoing transaction: neither as a pre- not a post-commit. The events for both transactional and non-transactional tables are saved in the transaction cache, and written to the binary log at end of the statement.
2006-12-21 09:29:02 +01:00
#define FLAGSTR(V,F) ((V)&(F)?#F" ":"")
Refactoring: Moved rli code to new file rpl_rli.cc. The idea being that rli should be a separate module (i.e. a class) to make it easier to maintain the code, e.g. by having checks within the rli checking sanity of data and making member variables private. This will also ease implementation of multi-source and, at least in my fantasies :), make it possible in some future to have separate replication servers.
2006-10-31 12:23:14 +01:00
"After Monty's review" changes to the fix for BUG#8325 "Deadlock in replication thread stops replication": s/sleep/safe_sleep (thread safe); sleep 0/1/2/3/4/5/5/5 (get slave less late); no message on error log (deadlock is too common sometimes), a global counter instead (SHOW STATUS LIKE 'slave_retried_transactions'). Plus a fix for libmysql/Makefile.shared
2005-03-23 19:19:36 +01:00
#define MAX_SLAVE_RETRY_PAUSE 5
post-merge fixes, including slave-skip-errors backport fixed too quick timeout in mysql-test-run which caused a race with the new server getting started before the old one completely finished shutdown. This should fix the pid warning we've been getting as well as inconsistent results when running tests with the manager
2002-01-22 23:05:11 +01:00
bool use_slave_mask = 0;
MY_BITMAP slave_error_mask;
Bug#28234 - global/session scope - documentation vs implementation Several system variables did not behave like system variables should do. When trying to SET them or use them in SELECT, they were reported as "unknown system variable". But they appeared in SHOW VARIABLES. This has been fixed by removing the "fixed_vars" array of variables and integrating the variables into the normal system variables chain. All of these variables do now behave as read-only global-only variables. Trying to SET them tells they are read-only, trying to SELECT the session value tells they are global only. Selecting the global value works. It delivers the same value as SHOW VARIABLES.
2008-11-22 00:22:21 +01:00
char slave_skip_error_names[SHOW_VAR_FUNC_BUFF_SIZE];
post-merge fixes, including slave-skip-errors backport fixed too quick timeout in mysql-test-run which caused a race with the new server getting started before the old one completely finished shutdown. This should fix the pid warning we've been getting as well as inconsistent results when running tests with the manager
2002-01-22 23:05:11 +01:00
fixed replication assert failure ( found by Jeremy Zawodny)
2002-03-08 23:02:11 +01:00
typedef bool (*CHECK_KILLED_FUNC)(THD*,void*);
LOAD DATA INFILE is now replicated properly, except for cleanup on Stop event and bugs the test suite could not catch Did some big restructuring of binlog event classes - most important change is that now each event class has exec_event method and one does not need to modify slave core code to add a new event. Slave code is now much smaller and easier to read
2001-08-03 23:57:53 +02:00
char* slave_load_tmpdir = 0;
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
Master_info *active_mi= 0;
Fixed several boolean type options used in my_long_options[]. Some options were declared as 'bool', but since those are being handled in my_getopt.c, bool can be machine dependent. To make sure it works in all circumstances, the type should be my_bool for C (not C++) programs.
2005-10-04 18:52:12 +02:00
my_bool replicate_same_server_id;
Changed relay_log_space to ulonglong Removed compiler warnings.
2002-08-23 14:14:01 +02:00
ulonglong relay_log_space_limit = 0;
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
When slave thread exits, we need to remember the temporary tables so we
can re-use them on slave start.
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
TODO: move the vars below under Master_info
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
*/
Added last_errno and last_error to show slave status slave will skip the first SQL_SLAVE_SKIP_COUNTER events on start - this is needed to recover from broken replication when the next log event(s) should just be skipped. Rotate binlogs when max_binlog_size is reached
2001-01-24 17:15:34 +01:00
added loops and other cool stuff to mysqltest fixed slave to preserve temp tables on slave stop; slave start added abort-slave-event-count option to mysqld to test the above added a test case for temp tables with a contantly aborting slave removed warning in sql_parse.cc fixed thimble.cc to compile
2000-12-02 18:11:50 +01:00
int disconnect_slave_event_count = 0, abort_slave_event_count = 0;
LOAD DATA INFILE is now replicated properly, except for cleanup on Stop event and bugs the test suite could not catch Did some big restructuring of binlog event classes - most important change is that now each event class has exec_event method and one does not need to modify slave core code to add a new event. Slave code is now much smaller and easier to read
2001-08-03 23:57:53 +02:00
int events_till_abort = -1;
Import changeset
2000-07-31 21:29:14 +02:00
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
enum enum_slave_reconnect_actions
{
SLAVE_RECON_ACT_REG= 0,
SLAVE_RECON_ACT_DUMP= 1,
SLAVE_RECON_ACT_EVENT= 2,
SLAVE_RECON_ACT_MAX
};
enum enum_slave_reconnect_messages
{
SLAVE_RECON_MSG_WAIT= 0,
SLAVE_RECON_MSG_KILLED_WAITING= 1,
SLAVE_RECON_MSG_AFTER= 2,
SLAVE_RECON_MSG_FAILED= 3,
SLAVE_RECON_MSG_COMMAND= 4,
SLAVE_RECON_MSG_KILLED_AFTER= 5,
SLAVE_RECON_MSG_MAX
};
static const char *reconnect_messages[SLAVE_RECON_ACT_MAX][SLAVE_RECON_MSG_MAX]=
{
{
"Waiting to reconnect after a failed registration on master",
"Slave I/O thread killed while waitnig to reconnect after a failed \
registration on master",
"Reconnecting after a failed registration on master",
"failed registering on master, reconnecting to try again, \
log '%s' at postion %s",
"COM_REGISTER_SLAVE",
"Slave I/O thread killed during or after reconnect"
},
{
"Waiting to reconnect after a failed binlog dump request",
"Slave I/O thread killed while retrying master dump",
"Reconnecting after a failed binlog dump request",
"failed dump request, reconnecting to try again, log '%s' at postion %s",
"COM_BINLOG_DUMP",
"Slave I/O thread killed during or after reconnect"
},
{
"Waiting to reconnect after a failed master event read",
"Slave I/O thread killed while waiting to reconnect after a failed read",
"Reconnecting after a failed master event read",
"Slave I/O thread: Failed reading log event, reconnecting to retry, \
log '%s' at postion %s",
"",
"Slave I/O thread killed during or after a reconnect done to recover from \
failed read"
}
};
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
typedef enum { SLAVE_THD_IO, SLAVE_THD_SQL} SLAVE_THD_TYPE;
fixed warning in sql_class.cc added gdb support, mysqld options per test, and other cool stuff to mysql-test-run added --replicate-wild-do/ignore-table more test cases
2000-11-21 07:38:08 +01:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int process_io_rotate(Master_info* mi, Rotate_log_event* rev);
static int process_io_create_file(Master_info* mi, Create_file_log_event* cev);
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
static bool wait_for_relay_log_space(Relay_log_info* rli);
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static inline bool io_slave_killed(THD* thd,Master_info* mi);
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
static inline bool sql_slave_killed(THD* thd,Relay_log_info* rli);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
static int init_slave_thread(THD* thd, SLAVE_THD_TYPE thd_type);
Bug#43835: SHOW VARIABLES does not include 0 for slave_skip_errors We didn't expect "error: no error", although this is in fact a legitimate state (if something is erroneous on the master, but not on the slave, e.g. INSERT fails on master due to UNIQUE constraint which does not exist on slave). We now list the ignore for "0: no error" the same way as any other ignore; moreover, if no or an empty --slave-skip-errors is passed at start-up, we show "OFF" instead of empty list, as intended. (The code for that was there, but was only run for the empty- argument case, even if it subsequently tested for both conditions.)
2009-04-06 13:42:33 +02:00
static void print_slave_skip_errors(void);
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int safe_connect(THD* thd, MYSQL* mysql, Master_info* mi);
static int safe_reconnect(THD* thd, MYSQL* mysql, Master_info* mi,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
bool suppress_warnings);
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int connect_to_master(THD* thd, MYSQL* mysql, Master_info* mi,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
bool reconnect, bool suppress_warnings);
fixed replication assert failure ( found by Jeremy Zawodny)
2002-03-08 23:02:11 +01:00
static int safe_sleep(THD* thd, int sec, CHECK_KILLED_FUNC thread_killed,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
void* thread_killed_arg);
updates for LOAD DATA FROM MASTER + some cleanup of replication code
2001-05-29 03:18:23 +02:00
static int request_table_dump(MYSQL* mysql, const char* db, const char* table);
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
static int create_table_from_dump(THD* thd, MYSQL *mysql, const char* db,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
const char* table_name, bool overwrite);
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int get_master_version_and_clock(MYSQL* mysql, Master_info* mi);
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
static Log_event* next_event(Relay_log_info* rli);
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
static int queue_event(Master_info* mi,const char* buf,ulong event_len);
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
static int terminate_slave_thread(THD *thd,
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
pthread_mutex_t *term_lock,
pthread_cond_t *term_cond,
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
volatile uint *slave_running,
bool skip_lock);
Fix for bug #29976: Excessive Slave I/O errors in replication tests Problem: The "Slave I/O thread couldn't register on master" error sporadically occurred in replication tests because the slave I/O thread got killed by STOP SLAVE before or while registering on master. Fixed by checking the state of the I/O thread, and issueing the error only if it was not explicitely killed by a user.
2007-11-22 17:22:54 +01:00
static bool check_io_slave_killed(THD *thd, Master_info *mi, const char *info);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
/*
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
Find out which replications threads are running
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
SYNOPSIS
init_thread_mask()
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mask Return value here
mi master_info for slave
inverse If set, returns which threads are not running
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
IMPLEMENTATION
Get a bit mask for which threads are running so that we can later restart
these threads.
RETURN
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mask If inverse == 0, running threads
If inverse == 1, stopped threads
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
void init_thread_mask(int* mask,Master_info* mi,bool inverse)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
bool set_io = mi->slave_running, set_sql = mi->rli.slave_running;
register int tmp_mask=0;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("init_thread_mask");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (set_io)
tmp_mask |= SLAVE_IO;
if (set_sql)
tmp_mask |= SLAVE_SQL;
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
if (inverse)
tmp_mask^= (SLAVE_IO | SLAVE_SQL);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
*mask = tmp_mask;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_VOID_RETURN;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
/*
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
lock_slave_threads()
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
*/
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
void lock_slave_threads(Master_info* mi)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("lock_slave_threads");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
//TODO: see if we can do this without dual mutex
pthread_mutex_lock(&mi->run_lock);
pthread_mutex_lock(&mi->rli.run_lock);
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_VOID_RETURN;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
/*
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
unlock_slave_threads()
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
*/
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
void unlock_slave_threads(Master_info* mi)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("unlock_slave_threads");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
//TODO: see if we can do this without dual mutex
pthread_mutex_unlock(&mi->rli.run_lock);
pthread_mutex_unlock(&mi->run_lock);
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_VOID_RETURN;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
/* Initialize slave structures */
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
int init_slave()
{
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
DBUG_ENTER("init_slave");
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Fix for BUG#2921 "Replication problem on mutex lock in mySQL-4.0.18": re-using unused LOCK_active_mi to serialize all administrative commands related to replication: START SLAVE, STOP SLAVE, RESET SLAVE, CHANGE MASTER, init_slave() (replication autostart at server startup), end_slave() (replication autostop at server shutdown), LOAD DATA FROM MASTER. This protects us against a handful of deadlocks (like BUG#2921 when two START SLAVE, but when two STOP SLAVE too). Removing unused variables.
2004-03-11 16:23:35 +01:00
/*
This is called when mysqld starts. Before client connections are
accepted. However bootstrap may conflict with us if it does START SLAVE.
So it's safer to take the lock.
*/
pthread_mutex_lock(&LOCK_active_mi);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
/*
TODO: re-write this to interate through the list of files
for multi-master
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
active_mi= new Master_info;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
Bug#43835: SHOW VARIABLES does not include 0 for slave_skip_errors We didn't expect "error: no error", although this is in fact a legitimate state (if something is erroneous on the master, but not on the slave, e.g. INSERT fails on master due to UNIQUE constraint which does not exist on slave). We now list the ignore for "0: no error" the same way as any other ignore; moreover, if no or an empty --slave-skip-errors is passed at start-up, we show "OFF" instead of empty list, as intended. (The code for that was there, but was only run for the empty- argument case, even if it subsequently tested for both conditions.)
2009-04-06 13:42:33 +02:00
/*
If --slave-skip-errors=... was not used, the string value for the
system variable has not been set up yet. Do it now.
*/
if (!use_slave_mask)
{
print_slave_skip_errors();
}
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
/*
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
If master_host is not specified, try to read it from the master_info file.
If master_host is specified, create the master_info file if it doesn't
exists.
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
*/
More error messages. This is intended to help debugging; presently I have a support issue with an unclear message which can have N reasons for appearing. This should help us know at which point it failed, and get the errno when my_open was involved (as the reason for the unclear message is often a permission problem). RESET SLAVE resets last_error and last_errno in SHOW SLAVE STATUS (without this, rpl_loaddata.test, which is expected to generate an error in last_error, influenced rpl_log_pos.test). A small test update. Added STOP SLAVE to mysql-test-run to get rid of several stupid error messages which are printed while the master restarts and the slave attempts/manages to connect to it and sends it nonsense binlog requests.
2003-06-10 23:29:49 +02:00
if (!active_mi)
{
sql_print_error("Failed to allocate memory for the master info structure");
goto err;
}
open binlog index file *before* tc_log->open() and binlog itself *after*
2005-02-01 15:36:48 +01:00
Simple code cleanup
2003-06-23 19:05:54 +02:00
if (init_master_info(active_mi,master_info_file,relay_log_info_file,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
!master_host, (SLAVE_IO | SLAVE_SQL)))
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
More error messages. This is intended to help debugging; presently I have a support issue with an unclear message which can have N reasons for appearing. This should help us know at which point it failed, and get the errno when my_open was involved (as the reason for the unclear message is often a permission problem). RESET SLAVE resets last_error and last_errno in SHOW SLAVE STATUS (without this, rpl_loaddata.test, which is expected to generate an error in last_error, influenced rpl_log_pos.test). A small test update. Added STOP SLAVE to mysql-test-run to get rid of several stupid error messages which are printed while the master restarts and the slave attempts/manages to connect to it and sends it nonsense binlog requests.
2003-06-10 23:29:49 +02:00
sql_print_error("Failed to initialize the master info structure");
Fixes for Netware Call pthread_mutex_destroy() on not used mutex. Changed comments in .h and .c files from // -> /* */ Added detection of mutex on which one didn't call pthread_mutex_destroy() Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression" Added optimisation for ORDER BY NULL
2003-01-28 07:38:28 +01:00
goto err;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
if (server_id && !master_host && active_mi->host[0])
master_host= active_mi->host;
Fix for BUG#3829 "Setting server_id on fly doesn't allow replication to start" (fix by our Harrison Fisk): when one does SET GLOBAL SERVER_ID=x, we must set server_id_supplied to 1.
2004-05-19 15:03:32 +02:00
/* If server id is not set, start_slave_thread() will say it */
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
if (master_host && !opt_skip_slave_start)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
if (start_slave_threads(1 /* need mutex */,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
0 /* no wait for start*/,
active_mi,
master_info_file,
relay_log_info_file,
SLAVE_IO | SLAVE_SQL))
Fixes for Netware Call pthread_mutex_destroy() on not used mutex. Changed comments in .h and .c files from // -> /* */ Added detection of mutex on which one didn't call pthread_mutex_destroy() Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression" Added optimisation for ORDER BY NULL
2003-01-28 07:38:28 +01:00
{
More error messages. This is intended to help debugging; presently I have a support issue with an unclear message which can have N reasons for appearing. This should help us know at which point it failed, and get the errno when my_open was involved (as the reason for the unclear message is often a permission problem). RESET SLAVE resets last_error and last_errno in SHOW SLAVE STATUS (without this, rpl_loaddata.test, which is expected to generate an error in last_error, influenced rpl_log_pos.test). A small test update. Added STOP SLAVE to mysql-test-run to get rid of several stupid error messages which are printed while the master restarts and the slave attempts/manages to connect to it and sends it nonsense binlog requests.
2003-06-10 23:29:49 +02:00
sql_print_error("Failed to create slave threads");
Fixes for Netware Call pthread_mutex_destroy() on not used mutex. Changed comments in .h and .c files from // -> /* */ Added detection of mutex on which one didn't call pthread_mutex_destroy() Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression" Added optimisation for ORDER BY NULL
2003-01-28 07:38:28 +01:00
goto err;
}
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Fix for BUG#2921 "Replication problem on mutex lock in mySQL-4.0.18": re-using unused LOCK_active_mi to serialize all administrative commands related to replication: START SLAVE, STOP SLAVE, RESET SLAVE, CHANGE MASTER, init_slave() (replication autostart at server startup), end_slave() (replication autostop at server shutdown), LOAD DATA FROM MASTER. This protects us against a handful of deadlocks (like BUG#2921 when two START SLAVE, but when two STOP SLAVE too). Removing unused variables.
2004-03-11 16:23:35 +01:00
pthread_mutex_unlock(&LOCK_active_mi);
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
DBUG_RETURN(0);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
Fixes for Netware Call pthread_mutex_destroy() on not used mutex. Changed comments in .h and .c files from // -> /* */ Added detection of mutex on which one didn't call pthread_mutex_destroy() Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression" Added optimisation for ORDER BY NULL
2003-01-28 07:38:28 +01:00
err:
Fix for BUG#2921 "Replication problem on mutex lock in mySQL-4.0.18": re-using unused LOCK_active_mi to serialize all administrative commands related to replication: START SLAVE, STOP SLAVE, RESET SLAVE, CHANGE MASTER, init_slave() (replication autostart at server startup), end_slave() (replication autostop at server shutdown), LOAD DATA FROM MASTER. This protects us against a handful of deadlocks (like BUG#2921 when two START SLAVE, but when two STOP SLAVE too). Removing unused variables.
2004-03-11 16:23:35 +01:00
pthread_mutex_unlock(&LOCK_active_mi);
Fixes for Netware Call pthread_mutex_destroy() on not used mutex. Changed comments in .h and .c files from // -> /* */ Added detection of mutex on which one didn't call pthread_mutex_destroy() Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression" Added optimisation for ORDER BY NULL
2003-01-28 07:38:28 +01:00
DBUG_RETURN(1);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
Bug#28234 - global/session scope - documentation vs implementation Several system variables did not behave like system variables should do. When trying to SET them or use them in SELECT, they were reported as "unknown system variable". But they appeared in SHOW VARIABLES. This has been fixed by removing the "fixed_vars" array of variables and integrating the variables into the normal system variables chain. All of these variables do now behave as read-only global-only variables. Trying to SET them tells they are read-only, trying to SELECT the session value tells they are global only. Selecting the global value works. It delivers the same value as SHOW VARIABLES.
2008-11-22 00:22:21 +01:00
/**
Convert slave skip errors bitmap into a printable string.
*/
static void print_slave_skip_errors(void)
{
/*
To be safe, we want 10 characters of room in the buffer for a number
plus terminators. Also, we need some space for constant strings.
10 characters must be sufficient for a number plus {',' | '...'}
plus a NUL terminator. That is a max 6 digit number.
*/
Bug#28234 - global/session scope - documentation vs implementation Post-pushbuild fix. - Windows does not have 'socket' system variable. - Compiler warning in sql/slave.cc
2008-11-27 11:50:28 +01:00
const size_t MIN_ROOM= 10;
Bug#28234 - global/session scope - documentation vs implementation Several system variables did not behave like system variables should do. When trying to SET them or use them in SELECT, they were reported as "unknown system variable". But they appeared in SHOW VARIABLES. This has been fixed by removing the "fixed_vars" array of variables and integrating the variables into the normal system variables chain. All of these variables do now behave as read-only global-only variables. Trying to SET them tells they are read-only, trying to SELECT the session value tells they are global only. Selecting the global value works. It delivers the same value as SHOW VARIABLES.
2008-11-22 00:22:21 +01:00
DBUG_ENTER("print_slave_skip_errors");
DBUG_ASSERT(sizeof(slave_skip_error_names) > MIN_ROOM);
DBUG_ASSERT(MAX_SLAVE_ERROR <= 999999); // 6 digits
if (!use_slave_mask || bitmap_is_clear_all(&slave_error_mask))
{
/* purecov: begin tested */
memcpy(slave_skip_error_names, STRING_WITH_LEN("OFF"));
/* purecov: end */
}
else if (bitmap_is_set_all(&slave_error_mask))
{
/* purecov: begin tested */
memcpy(slave_skip_error_names, STRING_WITH_LEN("ALL"));
/* purecov: end */
}
else
{
char *buff= slave_skip_error_names;
char *bend= buff + sizeof(slave_skip_error_names);
int errnum;
Bug#43835: SHOW VARIABLES does not include 0 for slave_skip_errors We didn't expect "error: no error", although this is in fact a legitimate state (if something is erroneous on the master, but not on the slave, e.g. INSERT fails on master due to UNIQUE constraint which does not exist on slave). We now list the ignore for "0: no error" the same way as any other ignore; moreover, if no or an empty --slave-skip-errors is passed at start-up, we show "OFF" instead of empty list, as intended. (The code for that was there, but was only run for the empty- argument case, even if it subsequently tested for both conditions.)
2009-04-06 13:42:33 +02:00
for (errnum= 0; errnum < MAX_SLAVE_ERROR; errnum++)
Bug#28234 - global/session scope - documentation vs implementation Several system variables did not behave like system variables should do. When trying to SET them or use them in SELECT, they were reported as "unknown system variable". But they appeared in SHOW VARIABLES. This has been fixed by removing the "fixed_vars" array of variables and integrating the variables into the normal system variables chain. All of these variables do now behave as read-only global-only variables. Trying to SET them tells they are read-only, trying to SELECT the session value tells they are global only. Selecting the global value works. It delivers the same value as SHOW VARIABLES.
2008-11-22 00:22:21 +01:00
{
if (bitmap_is_set(&slave_error_mask, errnum))
{
if (buff + MIN_ROOM >= bend)
break; /* purecov: tested */
buff= int10_to_str(errnum, buff, 10);
*buff++= ',';
}
}
if (buff != slave_skip_error_names)
buff--; // Remove last ','
if (errnum < MAX_SLAVE_ERROR)
{
/* Couldn't show all errors */
buff= strmov(buff, "..."); /* purecov: tested */
}
*buff=0;
}
DBUG_PRINT("init", ("error_names: '%s'", slave_skip_error_names));
DBUG_VOID_RETURN;
}
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
/*
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
Init function to set up array for errors that should be skipped for slave
Portability fix
2002-04-30 15:40:46 +02:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
SYNOPSIS
init_slave_skip_errors()
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
arg List of errors numbers to skip, separated with ','
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
NOTES
Called from get_options() in mysqld.cc on start-up
*/
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Portability fix
2002-04-30 15:40:46 +02:00
void init_slave_skip_errors(const char* arg)
post-merge fixes, including slave-skip-errors backport fixed too quick timeout in mysql-test-run which caused a race with the new server getting started before the old one completely finished shutdown. This should fix the pid warning we've been getting as well as inconsistent results when running tests with the manager
2002-01-22 23:05:11 +01:00
{
Portability fix
2002-04-30 15:40:46 +02:00
const char *p;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("init_slave_skip_errors");
key_map with more that 64 bits width
2003-10-11 13:06:55 +02:00
if (bitmap_init(&slave_error_mask,0,MAX_SLAVE_ERROR,0))
post-merge fixes, including slave-skip-errors backport fixed too quick timeout in mysql-test-run which caused a race with the new server getting started before the old one completely finished shutdown. This should fix the pid warning we've been getting as well as inconsistent results when running tests with the manager
2002-01-22 23:05:11 +01:00
{
fprintf(stderr, "Badly out of memory, please check your system status\n");
exit(1);
}
use_slave_mask = 1;
New ctype functions/macros to support many charsets at a time
2002-03-12 18:37:58 +01:00
for (;my_isspace(system_charset_info,*arg);++arg)
post-merge fixes, including slave-skip-errors backport fixed too quick timeout in mysql-test-run which caused a race with the new server getting started before the old one completely finished shutdown. This should fix the pid warning we've been getting as well as inconsistent results when running tests with the manager
2002-01-22 23:05:11 +01:00
/* empty */;
my_strncasecmp() is not used anymore. Use my_strncoll() instead.
2003-04-01 11:17:28 +02:00
if (!my_strnncoll(system_charset_info,(uchar*)arg,4,(const uchar*)"all",4))
post-merge fixes, including slave-skip-errors backport fixed too quick timeout in mysql-test-run which caused a race with the new server getting started before the old one completely finished shutdown. This should fix the pid warning we've been getting as well as inconsistent results when running tests with the manager
2002-01-22 23:05:11 +01:00
{
bitmap_set_all(&slave_error_mask);
BUG#39393 slave-skip-errors does not work when using ROW based replication RBR was not considering the option --slave-skip-errors. To fix the problem, we are reporting the ignored ERROR(s) as warnings thus avoiding stopping the SQL Thread. Besides, it fixes the output of "SHOW VARIABLES LIKE 'slave_skip_errors'" which was showing nothing when the value "all" was assigned to --slave-skip-errors. @sql/log_event.cc skipped rbr errors when the option skip-slave-errors is set. @sql/slave.cc fixed the output of for SHOW VARIABLES LIKE 'slave_skip_errors'" @test-cases fixed the output of rpl.rpl_idempotency updated the test case rpl_skip_error
2009-04-05 14:03:04 +02:00
print_slave_skip_errors();
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_VOID_RETURN;
post-merge fixes, including slave-skip-errors backport fixed too quick timeout in mysql-test-run which caused a race with the new server getting started before the old one completely finished shutdown. This should fix the pid warning we've been getting as well as inconsistent results when running tests with the manager
2002-01-22 23:05:11 +01:00
}
for (p= arg ; *p; )
{
long err_code;
if (!(p= str2int(p, 10, 0, LONG_MAX, &err_code)))
break;
if (err_code < MAX_SLAVE_ERROR)
bitmap_set_bit(&slave_error_mask,(uint)err_code);
New ctype functions/macros to support many charsets at a time
2002-03-12 18:37:58 +01:00
while (!my_isdigit(system_charset_info,*p) && *p)
post-merge fixes, including slave-skip-errors backport fixed too quick timeout in mysql-test-run which caused a race with the new server getting started before the old one completely finished shutdown. This should fix the pid warning we've been getting as well as inconsistent results when running tests with the manager
2002-01-22 23:05:11 +01:00
p++;
}
Bug#28234 - global/session scope - documentation vs implementation Several system variables did not behave like system variables should do. When trying to SET them or use them in SELECT, they were reported as "unknown system variable". But they appeared in SHOW VARIABLES. This has been fixed by removing the "fixed_vars" array of variables and integrating the variables into the normal system variables chain. All of these variables do now behave as read-only global-only variables. Trying to SET them tells they are read-only, trying to SELECT the session value tells they are global only. Selecting the global value works. It delivers the same value as SHOW VARIABLES.
2008-11-22 00:22:21 +01:00
/* Convert slave skip errors bitmap into a printable string. */
print_slave_skip_errors();
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_VOID_RETURN;
post-merge fixes, including slave-skip-errors backport fixed too quick timeout in mysql-test-run which caused a race with the new server getting started before the old one completely finished shutdown. This should fix the pid warning we've been getting as well as inconsistent results when running tests with the manager
2002-01-22 23:05:11 +01:00
}
BUG#41725: upmerge: 5.0-bt --> 5.1-bt
2009-05-23 01:29:41 +02:00
static void set_thd_in_use_temporary_tables(Relay_log_info *rli)
BUG#41725: slave crashes when inserting into temporary table after stop/start slave When stopping and restarting the slave while it is replicating temporary tables, the server would crash or raise an assertion failure. This was due to the fact that although temporary tables are saved between slave threads restart, the reference to the thread in use (table->in_use) was not being properly updated when the restart happened (it would still reference the old/invalid thread instead of the new one). This patch addresses this issue by resetting the reference to the new slave thread on slave thread restart.
2009-05-23 01:15:21 +02:00
{
TABLE *table;
for (table= rli->save_temporary_tables ; table ; table= table->next)
table->in_use= rli->sql_thd;
}
This is a fix for a bug in 3.23 -> 4.0 replication: Exec_master_log_pos is always too big by 6 bytes. So I add code to substract 6 bytes if the master is 3.23. This is not perfect (because it won't work if the slave I/O thread has not noticed yet that the master is 3.23), but as long as the slave I/O thread starts Exec_master_log_pos will be ok. It must be merged to 4.1 but not to 5.0 (or it can be, because of #if MYSQL_VERSION_ID), because 5.0 already works if the master is 3.23 (and in a more natural way: in 5.0 we store the end_log_pos in the binlog and relay log). I had to move functions from slave.h to slave.cc to satisfy gcc.
2004-04-07 00:57:14 +02:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
int terminate_slave_threads(Master_info* mi,int thread_mask,bool skip_lock)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("terminate_slave_threads");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (!mi->inited)
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0); /* successfully do nothing */
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
int error,force_all = (thread_mask & SLAVE_FORCE_ALL);
pthread_mutex_t *sql_lock = &mi->rli.run_lock, *io_lock = &mi->run_lock;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
if (thread_mask & (SLAVE_IO|SLAVE_FORCE_ALL))
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
Portability fixes Fixed test suite for HPUX 10.20 and MacOSX
2003-01-07 15:53:10 +01:00
DBUG_PRINT("info",("Terminating IO thread"));
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
mi->abort_slave=1;
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
if ((error=terminate_slave_thread(mi->io_thd, io_lock,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
&mi->stop_cond,
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
&mi->slave_running,
skip_lock)) &&
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
!force_all)
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
DBUG_RETURN(error);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
if (thread_mask & (SLAVE_SQL|SLAVE_FORCE_ALL))
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
Portability fixes Fixed test suite for HPUX 10.20 and MacOSX
2003-01-07 15:53:10 +01:00
DBUG_PRINT("info",("Terminating SQL thread"));
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
mi->rli.abort_slave=1;
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
if ((error=terminate_slave_thread(mi->rli.sql_thd, sql_lock,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
&mi->rli.stop_cond,
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
&mi->rli.slave_running,
skip_lock)) &&
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
!force_all)
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
DBUG_RETURN(error);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
DBUG_RETURN(0);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
/**
Wait for a slave thread to terminate.
This function is called after requesting the thread to terminate
(by setting @c abort_slave member of @c Relay_log_info or @c
Master_info structure to 1). Termination of the thread is
controlled with the the predicate <code>*slave_running</code>.
Function will acquire @c term_lock before waiting on the condition
unless @c skip_lock is true in which case the mutex should be owned
by the caller of this function and will remain acquired after
return from the function.
@param term_lock
Associated lock to use when waiting for @c term_cond
@param term_cond
Condition that is signalled when the thread has terminated
@param slave_running
Pointer to predicate to check for slave thread termination
@param skip_lock
If @c true the lock will not be acquired before waiting on
the condition. In this case, it is assumed that the calling
function acquires the lock before calling this function.
merging 5.0-bt to 5.1-bt
2009-04-30 15:17:46 +02:00
@retval 0 All OK ER_SLAVE_NOT_RUNNING otherwise.
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
merging 5.0-bt to 5.1-bt
2009-04-30 15:17:46 +02:00
@note If the executing thread has to acquire term_lock (skip_lock
is false), the negative running status does not represent
any issue therefore no error is reported.
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
*/
static int
terminate_slave_thread(THD *thd,
merging 5.0-bt to 5.1-bt
2009-04-30 15:17:46 +02:00
pthread_mutex_t *term_lock,
pthread_cond_t *term_cond,
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
volatile uint *slave_running,
bool skip_lock)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
WL#1062 "log charset info into all Query_log_event": we store 7 bytes (1 + 2*3) in every Query_log_event. In the future if users want binlog optimized for small size and less safe, we could add --binlog-no-charset (and binlog-no-sql-mode etc): charset info is something by design optional (even if for now we don't offer possibility to disable it): it's not a binlog format change. We try to reduce the number of get_charset() calls in the slave SQL thread to a minimum by caching the charset read from the previous event (which will often be equal to the one of the current event). We don't use SET ONE_SHOT for charset-aware repl (we still do for timezones, will be fixed later). No more errors if one changes the global value of charset vars on master or slave (as we log charset info in all Query_log_event). Not fixing Load_log_event as it will be rewritten soon by Dmitri. Testing how mysqlbinlog behaves in rpl_charset.test. mysqlbinlog needs to know where charset file is (to be able to convert a charset number found in binlog (e.g. in User_var_log_event) to a charset name); mysql-test-run needs to pass the correct value for this option to mysqlbinlog. Many result udpates (adding charset info into every event shifts log_pos in SHOW BINLOG EVENTS). Roughly the same job is to be done for timezones :)
2005-02-03 16:22:16 +01:00
DBUG_ENTER("terminate_slave_thread");
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
if (!skip_lock)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
pthread_mutex_lock(term_lock);
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
}
else
{
safe_mutex_assert_owner(term_lock);
}
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
if (!*slave_running)
{
if (!skip_lock)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
/*
if run_lock (term_lock) is acquired locally then either
slave_running status is fine
*/
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_mutex_unlock(term_lock);
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
DBUG_RETURN(0);
}
else
{
WL#1062 "log charset info into all Query_log_event": we store 7 bytes (1 + 2*3) in every Query_log_event. In the future if users want binlog optimized for small size and less safe, we could add --binlog-no-charset (and binlog-no-sql-mode etc): charset info is something by design optional (even if for now we don't offer possibility to disable it): it's not a binlog format change. We try to reduce the number of get_charset() calls in the slave SQL thread to a minimum by caching the charset read from the previous event (which will often be equal to the one of the current event). We don't use SET ONE_SHOT for charset-aware repl (we still do for timezones, will be fixed later). No more errors if one changes the global value of charset vars on master or slave (as we log charset info in all Query_log_event). Not fixing Load_log_event as it will be rewritten soon by Dmitri. Testing how mysqlbinlog behaves in rpl_charset.test. mysqlbinlog needs to know where charset file is (to be able to convert a charset number found in binlog (e.g. in User_var_log_event) to a charset name); mysql-test-run needs to pass the correct value for this option to mysqlbinlog. Many result udpates (adding charset info into every event shifts log_pos in SHOW BINLOG EVENTS). Roughly the same job is to be done for timezones :)
2005-02-03 16:22:16 +01:00
DBUG_RETURN(ER_SLAVE_NOT_RUNNING);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
}
DBUG_ASSERT(thd != 0);
auto-ROLLBACK if binlog was not closed properly auto-commit on Xid_log_event
2005-02-09 20:04:28 +01:00
THD_CHECK_SENTRY(thd);
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
Added the following new privleges: SHOW DATABASES CREATE TEMPORARY TABLE LOCK TABLES REPLICATION SLAVE & REPLICATION CLIENT SUPER EXECUTE All scripts & documentation is updated for this change. Added better error messages for global privileges
2002-06-12 14:04:18 +02:00
/*
auto-ROLLBACK if binlog was not closed properly auto-commit on Xid_log_event
2005-02-09 20:04:28 +01:00
Is is critical to test if the slave is running. Otherwise, we might
Added the following new privleges: SHOW DATABASES CREATE TEMPORARY TABLE LOCK TABLES REPLICATION SLAVE & REPLICATION CLIENT SUPER EXECUTE All scripts & documentation is updated for this change. Added better error messages for global privileges
2002-06-12 14:04:18 +02:00
be referening freed memory trying to kick it
replication updates and bugfixes. Still not perfect - there is some strange memory corruption I can only repeat on one system.
2002-03-16 02:44:44 +01:00
*/
Simple fixes (nothing notable)
2003-03-12 00:40:06 +01:00
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
while (*slave_running) // Should always be true
replication updates and bugfixes. Still not perfect - there is some strange memory corruption I can only repeat on one system.
2002-03-16 02:44:44 +01:00
{
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
int error;
WL#1062 "log charset info into all Query_log_event": we store 7 bytes (1 + 2*3) in every Query_log_event. In the future if users want binlog optimized for small size and less safe, we could add --binlog-no-charset (and binlog-no-sql-mode etc): charset info is something by design optional (even if for now we don't offer possibility to disable it): it's not a binlog format change. We try to reduce the number of get_charset() calls in the slave SQL thread to a minimum by caching the charset read from the previous event (which will often be equal to the one of the current event). We don't use SET ONE_SHOT for charset-aware repl (we still do for timezones, will be fixed later). No more errors if one changes the global value of charset vars on master or slave (as we log charset info in all Query_log_event). Not fixing Load_log_event as it will be rewritten soon by Dmitri. Testing how mysqlbinlog behaves in rpl_charset.test. mysqlbinlog needs to know where charset file is (to be able to convert a charset number found in binlog (e.g. in User_var_log_event) to a charset name); mysql-test-run needs to pass the correct value for this option to mysqlbinlog. Many result udpates (adding charset info into every event shifts log_pos in SHOW BINLOG EVENTS). Roughly the same job is to be done for timezones :)
2005-02-03 16:22:16 +01:00
DBUG_PRINT("loop", ("killing slave thread"));
BUG#31024: STOP SLAVE does not stop attempted connect()s Problem: if the IO slave thread is attempting to connect, STOP SLAVE waits for the attempt to finish. It may take a long time. Fix: don't wait, stop the slave immediately.
2008-03-11 14:42:54 +01:00
Bug #38816: kill + flush tables with read lock + stored procedures causes crashes! The problem of that bugreport was mostly fixed by the patch for bug 38691. However, attached test case focused on another crash or valgrind warning problem: SHOW PROCESSLIST query accesses freed memory of SP instruction that run in a parallel connection. Changes of thd->query/thd->query_length in dangerous places have been guarded with the per-thread LOCK_thd_data mutex (the THD::LOCK_delete mutex has been renamed to THD::LOCK_thd_data).
2009-07-24 17:58:58 +02:00
pthread_mutex_lock(&thd->LOCK_thd_data);
BUG#31024: STOP SLAVE does not stop attempted connect()s Problem: if the IO slave thread is attempting to connect, STOP SLAVE waits for the attempt to finish. It may take a long time. Fix: don't wait, stop the slave immediately.
2008-03-11 14:42:54 +01:00
#ifndef DONT_USE_THR_ALARM
/*
Error codes from pthread_kill are:
EINVAL: invalid signal number (can't happen)
ESRCH: thread already killed (can happen, should be ignored)
*/
IF_DBUG(int err= ) pthread_kill(thd->real_id, thr_client_alarm);
DBUG_ASSERT(err != EINVAL);
#endif
thd->awake(THD::NOT_KILLED);
Bug #38816: kill + flush tables with read lock + stored procedures causes crashes! The problem of that bugreport was mostly fixed by the patch for bug 38691. However, attached test case focused on another crash or valgrind warning problem: SHOW PROCESSLIST query accesses freed memory of SP instruction that run in a parallel connection. Changes of thd->query/thd->query_length in dangerous places have been guarded with the per-thread LOCK_thd_data mutex (the THD::LOCK_delete mutex has been renamed to THD::LOCK_thd_data).
2009-07-24 17:58:58 +02:00
pthread_mutex_unlock(&thd->LOCK_thd_data);
BUG#31024: STOP SLAVE does not stop attempted connect()s Problem: if the IO slave thread is attempting to connect, STOP SLAVE waits for the attempt to finish. It may take a long time. Fix: don't wait, stop the slave immediately.
2008-03-11 14:42:54 +01:00
Added the following new privleges: SHOW DATABASES CREATE TEMPORARY TABLE LOCK TABLES REPLICATION SLAVE & REPLICATION CLIENT SUPER EXECUTE All scripts & documentation is updated for this change. Added better error messages for global privileges
2002-06-12 14:04:18 +02:00
/*
There is a small chance that slave thread might miss the first
alarm. To protect againts it, resend the signal until it reacts
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
*/
struct timespec abstime;
Added the following new privleges: SHOW DATABASES CREATE TEMPORARY TABLE LOCK TABLES REPLICATION SLAVE & REPLICATION CLIENT SUPER EXECUTE All scripts & documentation is updated for this change. Added better error messages for global privileges
2002-06-12 14:04:18 +02:00
set_timespec(abstime,2);
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
error= pthread_cond_timedwait(term_cond, term_lock, &abstime);
DBUG_ASSERT(error == ETIMEDOUT || error == 0);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
BUG#29968 (rpl_ndb_circular.test and rpl_ndb_log.test fail): Removing unguarded read of slave_running field from inside terminate_slave_threads(). This could cause premature exit in the event that the slave thread already were shutting down, but isn't finished yet. The fields slave_running, io_thd, and sql_thread are guarded by an associated run_lock. A read of these fields were not guarded inside terminate_slave_threads(), which caused an assertion to fire. The assertion was removed, and the code reorganized slightly.
2007-08-29 16:06:59 +02:00
DBUG_ASSERT(*slave_running == 0);
if (!skip_lock)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_mutex_unlock(term_lock);
WL#1062 "log charset info into all Query_log_event": we store 7 bytes (1 + 2*3) in every Query_log_event. In the future if users want binlog optimized for small size and less safe, we could add --binlog-no-charset (and binlog-no-sql-mode etc): charset info is something by design optional (even if for now we don't offer possibility to disable it): it's not a binlog format change. We try to reduce the number of get_charset() calls in the slave SQL thread to a minimum by caching the charset read from the previous event (which will often be equal to the one of the current event). We don't use SET ONE_SHOT for charset-aware repl (we still do for timezones, will be fixed later). No more errors if one changes the global value of charset vars on master or slave (as we log charset info in all Query_log_event). Not fixing Load_log_event as it will be rewritten soon by Dmitri. Testing how mysqlbinlog behaves in rpl_charset.test. mysqlbinlog needs to know where charset file is (to be able to convert a charset number found in binlog (e.g. in User_var_log_event) to a charset name); mysql-test-run needs to pass the correct value for this option to mysqlbinlog. Many result udpates (adding charset info into every event shifts log_pos in SHOW BINLOG EVENTS). Roughly the same job is to be done for timezones :)
2005-02-03 16:22:16 +01:00
DBUG_RETURN(0);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
int start_slave_thread(pthread_handler h_func, pthread_mutex_t *start_lock,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
pthread_mutex_t *cond_lock,
pthread_cond_t *start_cond,
volatile uint *slave_running,
volatile ulong *slave_run_id,
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
Master_info* mi,
Replication: give the slave I/O thread a higher priority than the slave SQL thread so that it has fetched the most from the master's binlog before the master accidently dies. SCRUM task #756.
2003-03-01 23:59:27 +01:00
bool high_priority)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
pthread_t th;
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
ulong start_id;
DBUG_ENTER("start_slave_thread");
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ASSERT(mi->inited);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (start_lock)
pthread_mutex_lock(start_lock);
if (!server_id)
{
if (start_cond)
pthread_cond_broadcast(start_cond);
if (start_lock)
pthread_mutex_unlock(start_lock);
sql_print_error("Server id not set, will not start slave");
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
DBUG_RETURN(ER_BAD_SLAVE);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (*slave_running)
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
{
if (start_cond)
pthread_cond_broadcast(start_cond);
if (start_lock)
pthread_mutex_unlock(start_lock);
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
DBUG_RETURN(ER_SLAVE_MUST_STOP);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
}
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
start_id= *slave_run_id;
DBUG_PRINT("info",("Creating new slave thread"));
Replication: give the slave I/O thread a higher priority than the slave SQL thread so that it has fetched the most from the master's binlog before the master accidently dies. SCRUM task #756.
2003-03-01 23:59:27 +01:00
if (high_priority)
my_pthread_attr_setprio(&connection_attrib,CONNECT_PRIOR);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (pthread_create(&th, &connection_attrib, h_func, (void*)mi))
{
if (start_lock)
pthread_mutex_unlock(start_lock);
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
DBUG_RETURN(ER_SLAVE_THREAD);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Fix for: Bug #4810 "deadlock with KILL when the victim was in a wait state" (I included mutex unlock into exit_cond() for future safety) and BUG#4827 "KILL while START SLAVE may lead to replication slave crash"
2004-07-31 22:33:20 +02:00
if (start_cond && cond_lock) // caller has cond_lock
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
THD* thd = current_thd;
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
while (start_id == *slave_run_id)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
DBUG_PRINT("sleep",("Waiting for slave thread to start"));
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
const char* old_msg = thd->enter_cond(start_cond,cond_lock,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
"Waiting for slave thread to start");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_cond_wait(start_cond,cond_lock);
thd->exit_cond(old_msg);
Fix for: Bug #4810 "deadlock with KILL when the victim was in a wait state" (I included mutex unlock into exit_cond() for future safety) and BUG#4827 "KILL while START SLAVE may lead to replication slave crash"
2004-07-31 22:33:20 +02:00
pthread_mutex_lock(cond_lock); // re-acquire it as exit_cond() released
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (thd->killed)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
DBUG_RETURN(thd->killed_errno());
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
}
if (start_lock)
pthread_mutex_unlock(start_lock);
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
DBUG_RETURN(0);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
start_slave_threads()
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
NOTES
SLAVE_FORCE_ALL is not implemented here on purpose since it does not make
sense to do that for starting a slave--we always care if it actually
started the threads that were not previously running
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
*/
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
int start_slave_threads(bool need_slave_mutex, bool wait_for_start,
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
Master_info* mi, const char* master_info_fname,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
const char* slave_info_fname, int thread_mask)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
pthread_mutex_t *lock_io=0,*lock_sql=0,*lock_cond_io=0,*lock_cond_sql=0;
pthread_cond_t* cond_io=0,*cond_sql=0;
int error=0;
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
DBUG_ENTER("start_slave_threads");
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (need_slave_mutex)
{
lock_io = &mi->run_lock;
lock_sql = &mi->rli.run_lock;
}
if (wait_for_start)
{
cond_io = &mi->start_cond;
cond_sql = &mi->rli.start_cond;
lock_cond_io = &mi->run_lock;
lock_cond_sql = &mi->rli.run_lock;
}
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
if (thread_mask & SLAVE_IO)
error=start_slave_thread(handle_slave_io,lock_io,lock_cond_io,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
cond_io,
&mi->slave_running, &mi->slave_run_id,
mi, 1); //high priority, to read the most possible
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
if (!error && (thread_mask & SLAVE_SQL))
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
{
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
error=start_slave_thread(handle_slave_sql,lock_sql,lock_cond_sql,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
cond_sql,
&mi->rli.slave_running, &mi->rli.slave_run_id,
mi, 0);
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
if (error)
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
terminate_slave_threads(mi, thread_mask & SLAVE_IO, !need_slave_mutex);
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
}
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
DBUG_RETURN(error);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
laptop commit, syncing with the repostitory. Fixed some bad bugs in replication
2000-11-11 22:50:39 +01:00
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
#ifdef NOT_USED_YET
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int end_slave_on_walk(Master_info* mi, uchar* /*unused*/)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("end_slave_on_walk");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
end_master_info(mi);
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Changed relay_log_space to ulonglong Removed compiler warnings.
2002-08-23 14:14:01 +02:00
#endif
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
/*
Bug #38240 Crash in safe_mutex_lock () thr_mutex.c line 97 on rotate_relay_log The reason for the crash was rotate_relay_log (mi=0x0) did not verify the passed value of active_mi. There are more cases where active_mi is supposed to be non-zero e.g change_master(), stop_slave(), and it's reasonable to protect from a similar crash all of them with common fixes. Fixed with spliting end_slave() in slave threads release and slave data clean-up parts (a new close_active_mi()). The new function is invoked at the very end of close_connections() so that all users of active_mi are proven to have left.
2009-06-23 11:10:04 +02:00
Release slave threads at time of executing shutdown.
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
SYNOPSIS
end_slave()
*/
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
fixed up leaks found by --exit-info=256 better error diagnostic in SLAVE START
2000-12-09 22:28:51 +01:00
void end_slave()
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("end_slave");
Fix for BUG#2921 "Replication problem on mutex lock in mySQL-4.0.18": re-using unused LOCK_active_mi to serialize all administrative commands related to replication: START SLAVE, STOP SLAVE, RESET SLAVE, CHANGE MASTER, init_slave() (replication autostart at server startup), end_slave() (replication autostop at server shutdown), LOAD DATA FROM MASTER. This protects us against a handful of deadlocks (like BUG#2921 when two START SLAVE, but when two STOP SLAVE too). Removing unused variables.
2004-03-11 16:23:35 +01:00
/*
This is called when the server terminates, in close_connections().
It terminates slave threads. However, some CHANGE MASTER etc may still be
running presently. If a START SLAVE was in progress, the mutex lock below
will make us wait until slave threads have started, and START SLAVE
returns, then we terminate them here.
*/
pthread_mutex_lock(&LOCK_active_mi);
Fixes for Netware Call pthread_mutex_destroy() on not used mutex. Changed comments in .h and .c files from // -> /* */ Added detection of mutex on which one didn't call pthread_mutex_destroy() Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression" Added optimisation for ORDER BY NULL
2003-01-28 07:38:28 +01:00
if (active_mi)
{
/*
TODO: replace the line below with
list_walk(&master_list, (list_walk_action)end_slave_on_walk,0);
once multi-master code is ready.
*/
terminate_slave_threads(active_mi,SLAVE_FORCE_ALL);
Bug #38240 Crash in safe_mutex_lock () thr_mutex.c line 97 on rotate_relay_log The reason for the crash was rotate_relay_log (mi=0x0) did not verify the passed value of active_mi. There are more cases where active_mi is supposed to be non-zero e.g change_master(), stop_slave(), and it's reasonable to protect from a similar crash all of them with common fixes. Fixed with spliting end_slave() in slave threads release and slave data clean-up parts (a new close_active_mi()). The new function is invoked at the very end of close_connections() so that all users of active_mi are proven to have left.
2009-06-23 11:10:04 +02:00
}
pthread_mutex_unlock(&LOCK_active_mi);
DBUG_VOID_RETURN;
}
/**
Free all resources used by slave threads at time of executing shutdown.
The routine must be called after all possible users of @c active_mi
have left.
SYNOPSIS
close_active_mi()
*/
void close_active_mi()
{
pthread_mutex_lock(&LOCK_active_mi);
if (active_mi)
{
Fixes for Netware Call pthread_mutex_destroy() on not used mutex. Changed comments in .h and .c files from // -> /* */ Added detection of mutex on which one didn't call pthread_mutex_destroy() Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression" Added optimisation for ORDER BY NULL
2003-01-28 07:38:28 +01:00
end_master_info(active_mi);
delete active_mi;
active_mi= 0;
}
Fix for BUG#2921 "Replication problem on mutex lock in mySQL-4.0.18": re-using unused LOCK_active_mi to serialize all administrative commands related to replication: START SLAVE, STOP SLAVE, RESET SLAVE, CHANGE MASTER, init_slave() (replication autostart at server startup), end_slave() (replication autostop at server shutdown), LOAD DATA FROM MASTER. This protects us against a handful of deadlocks (like BUG#2921 when two START SLAVE, but when two STOP SLAVE too). Removing unused variables.
2004-03-11 16:23:35 +01:00
pthread_mutex_unlock(&LOCK_active_mi);
fixed up leaks found by --exit-info=256 better error diagnostic in SLAVE START
2000-12-09 22:28:51 +01:00
}
fixed warning in sql_class.cc added gdb support, mysqld options per test, and other cool stuff to mysql-test-run added --replicate-wild-do/ignore-table more test cases
2000-11-21 07:38:08 +01:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static bool io_slave_killed(THD* thd, Master_info* mi)
Import changeset
2000-07-31 21:29:14 +02:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("io_slave_killed");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(mi->io_thd == thd);
A change of behaviour of Seconds_Behind_Master from SHOW SLAVE STATUS. It's going into 4.1 because old behaviour was somewhat nonsensical (kind of bug). Changes are that if repl threads are down or disconnected the column will be NULL, and if master is idle the column will not grow indefinitely anymore.
2004-12-16 18:12:22 +01:00
DBUG_ASSERT(mi->slave_running); // tracking buffer overrun
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(mi->abort_slave || abort_loop || thd->killed);
Import changeset
2000-07-31 21:29:14 +02:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
static bool sql_slave_killed(THD* thd, Relay_log_info* rli)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("sql_slave_killed");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(rli->sql_thd == thd);
DBUG_ASSERT(rli->slave_running == 1);// tracking buffer overrun
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
if (abort_loop || thd->killed || rli->abort_slave)
{
Bug#38205 Row-based Replication (RBR) causes inconsistencies: HA_ERR_FOUND_DUP Bug#319 if while a non-transactional slave is replicating a transaction possible problem It is impossible to roll back a mixed engines transaction when one of the engine is non-transaction. In replication that fact is crucial because the slave can not safely re-apply a transction that was interrupted with STOP SLAVE. Fixed with making STOP SLAVE not be effective immediately in the case the current group of replication events has modified a non-transaction table. In order for slave to leave either the group needs finishing or the user issues KILL QUERY|CONNECTION slave_thread_id.
2009-03-26 09:25:06 +01:00
if (rli->abort_slave && rli->is_in_group() &&
thd->transaction.all.modified_non_trans_table)
DBUG_RETURN(0);
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
/*
If we are in an unsafe situation (stopping could corrupt replication),
we give one minute to the slave SQL thread of grace before really
terminating, in the hope that it will be able to read more events and
the unsafe situation will soon be left. Note that this one minute starts
from the last time anything happened in the slave SQL thread. So it's
really one minute of idleness, we don't timeout if the slave SQL thread
is actively working.
*/
BUG#23171 (Illegal slave restart position): Third patch of the bug fix where the code for skipping events and for executing events is factored out into three functions: - shall_skip() to decide if the event shall be skipped and the reason for it; - do_apply_event(), where the event is applied to the database; and - do_update_pos(), which updates the actual relay log position and group positions.
2007-01-17 15:06:37 +01:00
if (rli->last_event_start_time == 0)
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
DBUG_PRINT("info", ("Slave SQL thread is in an unsafe situation, giving "
"it some grace period"));
BUG#23171 (Illegal slave restart position): Third patch of the bug fix where the code for skipping events and for executing events is factored out into three functions: - shall_skip() to decide if the event shall be skipped and the reason for it; - do_apply_event(), where the event is applied to the database; and - do_update_pos(), which updates the actual relay log position and group positions.
2007-01-17 15:06:37 +01:00
if (difftime(time(0), rli->last_event_start_time) > 60)
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
{
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
rli->report(ERROR_LEVEL, 0,
"SQL thread had to stop in an unsafe situation, in "
"the middle of applying updates to a "
"non-transactional table without any primary key. "
"There is a risk of duplicate updates when the slave "
"SQL thread is restarted. Please check your tables' "
"contents after restart.");
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
}
}
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
/*
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
skip_load_data_infile()
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
NOTES
This is used to tell a 3.23 master to break send_file()
*/
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
void skip_load_data_infile(NET *net)
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("skip_load_data_infile");
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
(void)net_request_file(net, "/dev/null");
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
(void)my_net_read(net); // discard response
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
(void)net_write_command(net, 0, (uchar*) "", 0, (uchar*) "", 0); // ok
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_VOID_RETURN;
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
bool net_request_file(NET* net, const char* fname)
skip updates with the same server id as self kill zombie COM_BINLOG_DUMP with the same server id on connect
2000-10-03 01:59:12 +02:00
{
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
DBUG_ENTER("net_request_file");
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
DBUG_RETURN(net_write_command(net, 251, (uchar*) fname, strlen(fname),
(uchar*) "", 0));
skip updates with the same server id as self kill zombie COM_BINLOG_DUMP with the same server id on connect
2000-10-03 01:59:12 +02:00
}
WL#1036 (print the db in slave error messages). I extended the task to cleaning error messages, making them look nicer, and making the output of SHOW SLAVE STATUS (column Last_error) be as complete as what's printed on the .err file; previously we would have, for a failure of a replicated LOAD DATA INFILE: - in the .err, 2 lines: "duplicate entry 2708 for key 1" "failed loading SQL_LOAD-5-2-2.info" - and in SHOW SLAVE STATUS, only: "failed loading SQL_LOAD-5-2-2.info". Now SHOW SLAVE STATUS will contain the concatenation of the 2 messages.
2003-07-24 22:25:36 +02:00
/*
From other comments and tests in code, it looks like
Review of changesets since last pull. (Mostly code cleanups)
2003-08-07 19:16:37 +02:00
sometimes Query_log_event and Load_log_event can have db == 0
WL#1036 (print the db in slave error messages). I extended the task to cleaning error messages, making them look nicer, and making the output of SHOW SLAVE STATUS (column Last_error) be as complete as what's printed on the .err file; previously we would have, for a failure of a replicated LOAD DATA INFILE: - in the .err, 2 lines: "duplicate entry 2708 for key 1" "failed loading SQL_LOAD-5-2-2.info" - and in SHOW SLAVE STATUS, only: "failed loading SQL_LOAD-5-2-2.info". Now SHOW SLAVE STATUS will contain the concatenation of the 2 messages.
2003-07-24 22:25:36 +02:00
(see rewrite_db() above for example)
(cases where this happens are unclear; it may be when the master is 3.23).
*/
Review of changesets since last pull. (Mostly code cleanups)
2003-08-07 19:16:37 +02:00
const char *print_slave_db_safe(const char* db)
WL#1036 (print the db in slave error messages). I extended the task to cleaning error messages, making them look nicer, and making the output of SHOW SLAVE STATUS (column Last_error) be as complete as what's printed on the .err file; previously we would have, for a failure of a replicated LOAD DATA INFILE: - in the .err, 2 lines: "duplicate entry 2708 for key 1" "failed loading SQL_LOAD-5-2-2.info" - and in SHOW SLAVE STATUS, only: "failed loading SQL_LOAD-5-2-2.info". Now SHOW SLAVE STATUS will contain the concatenation of the 2 messages.
2003-07-24 22:25:36 +02:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("*print_slave_db_safe");
DBUG_RETURN((db ? db : ""));
WL#1036 (print the db in slave error messages). I extended the task to cleaning error messages, making them look nicer, and making the output of SHOW SLAVE STATUS (column Last_error) be as complete as what's printed on the .err file; previously we would have, for a failure of a replicated LOAD DATA INFILE: - in the .err, 2 lines: "duplicate entry 2708 for key 1" "failed loading SQL_LOAD-5-2-2.info" - and in SHOW SLAVE STATUS, only: "failed loading SQL_LOAD-5-2-2.info". Now SHOW SLAVE STATUS will contain the concatenation of the 2 messages.
2003-07-24 22:25:36 +02:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Refactoring: Moved rli code to new file rpl_rli.cc. The idea being that rli should be a separate module (i.e. a class) to make it easier to maintain the code, e.g. by having checks within the rli checking sanity of data and making member variables private. This will also ease implementation of multi-source and, at least in my fantasies :), make it possible in some future to have separate replication servers.
2006-10-31 12:23:14 +01:00
int init_strvar_from_file(char *var, int max_size, IO_CACHE *f,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
const char *default_val)
Import changeset
2000-07-31 21:29:14 +02:00
{
keep bk happy
2001-01-22 03:46:32 +01:00
uint length;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("init_strvar_from_file");
keep bk happy
2001-01-22 03:46:32 +01:00
if ((length=my_b_gets(f,var, max_size)))
{
char* last_p = var + length -1;
if (*last_p == '\n')
*last_p = 0; // if we stopped on newline, kill it
else
Import changeset
2000-07-31 21:29:14 +02:00
{
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
If we truncated a line or stopped on last char, remove all chars
up to and including newline.
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
*/
keep bk happy
2001-01-22 03:46:32 +01:00
int c;
Bug #43414 Parenthesis (and other) warnings compiling MySQL with gcc 4.3.2 Compiling MySQL with gcc 4.3.2 and later produces a number of warnings, many of which are new with the recent compiler versions. This bug will be resolved in more than one patch to limit the size of changesets. This is the first patch, fixing a number of the warnings, predominantly "suggest using parentheses around && in ||", and empty for and while bodies.
2009-06-09 18:11:21 +02:00
while (((c=my_b_get(f)) != '\n' && c != my_b_EOF)) ;
client/mysqltest.c fixed bug that created empty result files in no-record mode mysql-test/t/rpl000007.test testing non-replication of load data infile when the table is excluded from replication mysql-test/t/rpl000016.test test purge master logs sql/slave.cc fix memory leaks found by purge master logs test sql/sql_repl.cc fixed race condition and delete before close bug in reset_slave() Overall comments - coverage testing is a very good thing!
2000-12-15 04:17:18 +01:00
}
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
keep bk happy
2001-01-22 03:46:32 +01:00
}
else if (default_val)
{
Fix for make dist Made replication test portable Fixed buffer overrun bug in replication
2001-10-02 21:21:14 +02:00
strmake(var, default_val, max_size-1);
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
keep bk happy
2001-01-22 03:46:32 +01:00
}
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
Import changeset
2000-07-31 21:29:14 +02:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Refactoring: Moved rli code to new file rpl_rli.cc. The idea being that rli should be a separate module (i.e. a class) to make it easier to maintain the code, e.g. by having checks within the rli checking sanity of data and making member variables private. This will also ease implementation of multi-source and, at least in my fantasies :), make it possible in some future to have separate replication servers.
2006-10-31 12:23:14 +01:00
int init_intvar_from_file(int* var, IO_CACHE* f, int default_val)
Import changeset
2000-07-31 21:29:14 +02:00
{
char buf[32];
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("init_intvar_from_file");
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
if (my_b_gets(f, buf, sizeof(buf)))
keep bk happy
2001-01-22 03:46:32 +01:00
{
*var = atoi(buf);
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
keep bk happy
2001-01-22 03:46:32 +01:00
}
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
else if (default_val)
keep bk happy
2001-01-22 03:46:32 +01:00
{
*var = default_val;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
keep bk happy
2001-01-22 03:46:32 +01:00
}
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
Import changeset
2000-07-31 21:29:14 +02:00
}
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
/*
Check if the error is caused by network.
@param[in] errorno Number of the error.
RETURNS:
TRUE network error
FALSE not network error
*/
bool is_network_error(uint errorno)
{
if (errorno == CR_CONNECTION_ERROR ||
errorno == CR_CONN_HOST_ERROR ||
errorno == CR_SERVER_GONE_ERROR ||
errorno == CR_SERVER_LOST ||
errorno == ER_CON_COUNT_ERROR ||
errorno == ER_SERVER_SHUTDOWN)
return TRUE;
return FALSE;
}
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
/*
Note that we rely on the master's version (3.23, 4.0.14 etc) instead of
relying on the binlog's version. This is not perfect: imagine an upgrade
of the master without waiting that all slaves are in sync with the master;
then a slave could be fooled about the binlog's format. This is what happens
when people upgrade a 3.23 master to 4.0 without doing RESET MASTER: 4.0
slaves are fooled. So we do this only to distinguish between 3.23 and more
recent masters (it's too late to change things for 3.23).
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
RETURNS
0 ok
1 error
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
2 transient network problem, the caller should try to reconnect
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
*/
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int get_master_version_and_clock(MYSQL* mysql, Master_info* mi)
work to enable reading 3.23 logs - not yet finished moved fail-safe replication routines from sql_repl.cc to repl_failsafe.cc write start event only in the first log
2001-11-11 06:24:12 +01:00
{
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
char err_buff[MAX_SLAVE_ERRMSG];
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
const char* errmsg= 0;
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
int err_code= 0;
MYSQL_RES *master_res= 0;
MYSQL_ROW master_row;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("get_master_version_and_clock");
This is the final commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) It contains a few bugfixes (which I made when running the testsuite). I carefully updated the results of the testsuite (i.e. I checked for every one, if the difference between .reject and .result could be explained). Apparently mysql-test-run --manager is broken in 4.1 and 5.0 currently, so I could neither run the few tests which require --manager, nor check that they pass nor modify their .result. But for builds, we don't run with --manager. Apart from --manager, the full testsuite passes, with Valgrind too (no errors). I'm going to push in the next minutes. Remains: update the manual. Note: by chance I saw that (in 4.1, in 5.0) rpl_get_lock fails when run alone; this is normal at it makes assumptions on thread ids. I will fix this one day in 4.1.
2003-12-19 22:40:23 +01:00
/*
Free old description_event_for_queue (that is needed if we are in
a reconnection).
*/
delete mi->rli.relay_log.description_event_for_queue;
mi->rli.relay_log.description_event_for_queue= 0;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
if (!my_isdigit(&my_charset_bin,*mysql->server_version))
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
{
Post-merge fixes.
2003-10-23 16:06:51 +02:00
errmsg = "Master reported unrecognized MySQL version";
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
err_code= ER_SLAVE_FATAL_ERROR;
sprintf(err_buff, ER(err_code), errmsg);
}
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
else
{
/*
Note the following switch will bug when we have MySQL branch 30 ;)
*/
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
switch (*mysql->server_version)
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
{
case '0':
case '1':
case '2':
errmsg = "Master reported unrecognized MySQL version";
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
err_code= ER_SLAVE_FATAL_ERROR;
sprintf(err_buff, ER(err_code), errmsg);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
break;
case '3':
mi->rli.relay_log.description_event_for_queue= new
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Format_description_log_event(1, mysql->server_version);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
break;
case '4':
mi->rli.relay_log.description_event_for_queue= new
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Format_description_log_event(3, mysql->server_version);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
break;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
default:
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
/*
Master is MySQL >=5.0. Give a default Format_desc event, so that we can
take the early steps (like tests for "is this a 3.23 master") which we
have to take before we receive the real master's Format_desc which will
override this one. Note that the Format_desc we create below is garbage
(it has the format of the *slave*); it's only good to help know if the
master is 3.23, 4.0, etc.
*/
mi->rli.relay_log.description_event_for_queue= new
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Format_description_log_event(4, mysql->server_version);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
break;
}
work to enable reading 3.23 logs - not yet finished moved fail-safe replication routines from sql_repl.cc to repl_failsafe.cc write start event only in the first log
2001-11-11 06:24:12 +01:00
}
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
/*
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
This does not mean that a 5.0 slave will be able to read a 6.0 master; but
as we don't know yet, we don't want to forbid this for now. If a 5.0 slave
can't read a 6.0 master, this will show up when the slave can't read some
events sent by the master, and there will be error messages.
*/
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
if (errmsg)
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
goto err;
This is the final commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) It contains a few bugfixes (which I made when running the testsuite). I carefully updated the results of the testsuite (i.e. I checked for every one, if the difference between .reject and .result could be explained). Apparently mysql-test-run --manager is broken in 4.1 and 5.0 currently, so I could neither run the few tests which require --manager, nor check that they pass nor modify their .result. But for builds, we don't run with --manager. Apart from --manager, the full testsuite passes, with Valgrind too (no errors). I'm going to push in the next minutes. Remains: update the manual. Note: by chance I saw that (in 4.1, in 5.0) rpl_get_lock fails when run alone; this is normal at it makes assumptions on thread ids. I will fix this one day in 4.1.
2003-12-19 22:40:23 +01:00
/* as we are here, we tried to allocate the event */
if (!mi->rli.relay_log.description_event_for_queue)
{
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
errmsg= "default Format_description_log_event";
err_code= ER_SLAVE_CREATE_EVENT_FAILURE;
sprintf(err_buff, ER(err_code), errmsg);
goto err;
This is the final commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) It contains a few bugfixes (which I made when running the testsuite). I carefully updated the results of the testsuite (i.e. I checked for every one, if the difference between .reject and .result could be explained). Apparently mysql-test-run --manager is broken in 4.1 and 5.0 currently, so I could neither run the few tests which require --manager, nor check that they pass nor modify their .result. But for builds, we don't run with --manager. Apart from --manager, the full testsuite passes, with Valgrind too (no errors). I'm going to push in the next minutes. Remains: update the manual. Note: by chance I saw that (in 4.1, in 5.0) rpl_get_lock fails when run alone; this is normal at it makes assumptions on thread ids. I will fix this one day in 4.1.
2003-12-19 22:40:23 +01:00
}
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
/*
Compare the master and slave's clock. Do not die if master's clock is
unavailable (very old master not supporting UNIX_TIMESTAMP()?).
*/
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
DBUG_SYNC_POINT("debug_lock.before_get_UNIX_TIMESTAMP", 10);
master_res= NULL;
Inefficient usage of String::append() fixed. Bad examples of usage of a string with its length fixed. The incorrect length in the trigger file configuration descriptor fixed (BUG#14090). A hook for unknown keys added to the parser to support old .TRG files.
2005-11-20 19:47:07 +01:00
if (!mysql_real_query(mysql, STRING_WITH_LEN("SELECT UNIX_TIMESTAMP()")) &&
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
(master_res= mysql_store_result(mysql)) &&
(master_row= mysql_fetch_row(master_res)))
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mi->clock_diff_with_master=
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
(long) (time((time_t*) 0) - strtoul(master_row[0], 0, 10));
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
}
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
else if (is_network_error(mysql_errno(mysql)))
{
mi->report(WARNING_LEVEL, mysql_errno(mysql),
"Get master clock failed with error: %s", mysql_error(mysql));
goto network_err;
}
else
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
{
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
mi->clock_diff_with_master= 0; /* The "most sensible" value */
Fixed federated and some replication tests to not stop slave until it's up and running. (Removes some warnings about UNIX_TIMESTAMP from the slave.err logs) Marked federated_server as a '--big-test' Change error in net_clear to 'Note', as it interfered with mysql-test-run.
2007-04-17 13:41:16 +02:00
sql_print_warning("\"SELECT UNIX_TIMESTAMP()\" failed on master, "
"do not trust column Seconds_Behind_Master of SHOW "
"SLAVE STATUS. Error: %s (%d)",
mysql_error(mysql), mysql_errno(mysql));
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
}
if (master_res)
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mysql_free_result(master_res);
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
master_res= NULL;
}
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
/*
Check that the master's server id and ours are different. Because if they
are equal (which can result from a simple copy of master's datadir to slave,
thus copying some my.cnf), replication will work but all events will be
skipped.
Do not die if SHOW VARIABLES LIKE 'SERVER_ID' fails on master (very old
master?).
Note: we could have put a @@SERVER_ID in the previous SELECT
UNIX_TIMESTAMP() instead, but this would not have worked on 3.23 masters.
*/
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
DBUG_SYNC_POINT("debug_lock.before_get_SERVER_ID", 10);
master_res= NULL;
master_row= NULL;
Inefficient usage of String::append() fixed. Bad examples of usage of a string with its length fixed. The incorrect length in the trigger file configuration descriptor fixed (BUG#14090). A hook for unknown keys added to the parser to support old .TRG files.
2005-11-20 19:47:07 +01:00
if (!mysql_real_query(mysql,
STRING_WITH_LEN("SHOW VARIABLES LIKE 'SERVER_ID'")) &&
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
(master_res= mysql_store_result(mysql)) &&
(master_row= mysql_fetch_row(master_res)))
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
{
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
if ((::server_id == strtoul(master_row[1], 0, 10)) &&
BUG#23171: Illegal group log position Tail fixes after re-applying patches to older version of clone.
2007-03-22 08:32:41 +01:00
!mi->rli.replicate_same_server_id)
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
{
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
errmsg= "The slave I/O thread stops because master and slave have equal \
MySQL server ids; these ids must be different for replication to work (or \
the --replicate-same-server-id option must be used on slave but this does \
not always make sense; please check the manual before using it).";
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
err_code= ER_SLAVE_FATAL_ERROR;
sprintf(err_buff, ER(err_code), errmsg);
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
goto err;
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
}
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
}
else if (mysql_errno(mysql))
{
if (is_network_error(mysql_errno(mysql)))
{
mi->report(WARNING_LEVEL, mysql_errno(mysql),
"Get master SERVER_ID failed with error: %s", mysql_error(mysql));
goto network_err;
}
/* Fatal error */
errmsg= "The slave I/O thread stops because a fatal error is encountered \
when it try to get the value of SERVER_ID variable from master.";
err_code= mysql_errno(mysql);
sprintf(err_buff, "%s Error: %s", errmsg, mysql_error(mysql));
goto err;
}
else if (!master_row && master_res)
{
mi->report(WARNING_LEVEL, ER_UNKNOWN_SYSTEM_VARIABLE,
"Unknown system variable 'SERVER_ID' on master, \
maybe it is a *VERY OLD MASTER*.");
}
if (master_res)
{
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
mysql_free_result(master_res);
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
master_res= NULL;
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
}
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
/*
Check that the master's global character_set_server and ours are the same.
Not fatal if query fails (old master?).
Fix for BUG#4678 "mysql-test-run fails on grant_cache": do not use '' as user in tests, because it picks the Unix login (which gives unexpected results if it is 'root') (such behaviour is a feature of mysql_real_connect(), see the manual).
2004-08-12 13:12:09 +02:00
Note that we don't check for equality of global character_set_client and
collation_connection (neither do we prevent their setting in
set_var.cc). That's because from what I (Guilhem) have tested, the global
values of these 2 are never used (new connections don't use them).
We don't test equality of global collation_database either as it's is
going to be deprecated (made read-only) in 4.1 very soon.
WL#1062 "log charset info into all Query_log_event": we store 7 bytes (1 + 2*3) in every Query_log_event. In the future if users want binlog optimized for small size and less safe, we could add --binlog-no-charset (and binlog-no-sql-mode etc): charset info is something by design optional (even if for now we don't offer possibility to disable it): it's not a binlog format change. We try to reduce the number of get_charset() calls in the slave SQL thread to a minimum by caching the charset read from the previous event (which will often be equal to the one of the current event). We don't use SET ONE_SHOT for charset-aware repl (we still do for timezones, will be fixed later). No more errors if one changes the global value of charset vars on master or slave (as we log charset info in all Query_log_event). Not fixing Load_log_event as it will be rewritten soon by Dmitri. Testing how mysqlbinlog behaves in rpl_charset.test. mysqlbinlog needs to know where charset file is (to be able to convert a charset number found in binlog (e.g. in User_var_log_event) to a charset name); mysql-test-run needs to pass the correct value for this option to mysqlbinlog. Many result udpates (adding charset info into every event shifts log_pos in SHOW BINLOG EVENTS). Roughly the same job is to be done for timezones :)
2005-02-03 16:22:16 +01:00
The test is only relevant if master < 5.0.3 (we'll test only if it's older
than the 5 branch; < 5.0.3 was alpha...), as >= 5.0.3 master stores
charset info in each binlog event.
We don't do it for 3.23 because masters <3.23.50 hang on
SELECT @@unknown_var (BUG#7965 - see changelog of 3.23.50). So finally we
test only if master is 4.x.
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
*/
WL#1062 "log charset info into all Query_log_event": we store 7 bytes (1 + 2*3) in every Query_log_event. In the future if users want binlog optimized for small size and less safe, we could add --binlog-no-charset (and binlog-no-sql-mode etc): charset info is something by design optional (even if for now we don't offer possibility to disable it): it's not a binlog format change. We try to reduce the number of get_charset() calls in the slave SQL thread to a minimum by caching the charset read from the previous event (which will often be equal to the one of the current event). We don't use SET ONE_SHOT for charset-aware repl (we still do for timezones, will be fixed later). No more errors if one changes the global value of charset vars on master or slave (as we log charset info in all Query_log_event). Not fixing Load_log_event as it will be rewritten soon by Dmitri. Testing how mysqlbinlog behaves in rpl_charset.test. mysqlbinlog needs to know where charset file is (to be able to convert a charset number found in binlog (e.g. in User_var_log_event) to a charset name); mysql-test-run needs to pass the correct value for this option to mysqlbinlog. Many result udpates (adding charset info into every event shifts log_pos in SHOW BINLOG EVENTS). Roughly the same job is to be done for timezones :)
2005-02-03 16:22:16 +01:00
/* redundant with rest of code but safer against later additions */
if (*mysql->server_version == '3')
Fix for BUG#7965 "Slave_IO_State Stuck at 'Checking Master Version'": Working around hang of master < 3.23.50 on SELECT @@unknown_var (to enable 3.23.49->4.1.10 replication)
2005-01-17 21:26:14 +01:00
goto err;
WL#1062 "log charset info into all Query_log_event": we store 7 bytes (1 + 2*3) in every Query_log_event. In the future if users want binlog optimized for small size and less safe, we could add --binlog-no-charset (and binlog-no-sql-mode etc): charset info is something by design optional (even if for now we don't offer possibility to disable it): it's not a binlog format change. We try to reduce the number of get_charset() calls in the slave SQL thread to a minimum by caching the charset read from the previous event (which will often be equal to the one of the current event). We don't use SET ONE_SHOT for charset-aware repl (we still do for timezones, will be fixed later). No more errors if one changes the global value of charset vars on master or slave (as we log charset info in all Query_log_event). Not fixing Load_log_event as it will be rewritten soon by Dmitri. Testing how mysqlbinlog behaves in rpl_charset.test. mysqlbinlog needs to know where charset file is (to be able to convert a charset number found in binlog (e.g. in User_var_log_event) to a charset name); mysql-test-run needs to pass the correct value for this option to mysqlbinlog. Many result udpates (adding charset info into every event shifts log_pos in SHOW BINLOG EVENTS). Roughly the same job is to be done for timezones :)
2005-02-03 16:22:16 +01:00
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
if (*mysql->server_version == '4')
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
{
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
master_res= NULL;
if (!mysql_real_query(mysql,
STRING_WITH_LEN("SELECT @@GLOBAL.COLLATION_SERVER")) &&
(master_res= mysql_store_result(mysql)) &&
(master_row= mysql_fetch_row(master_res)))
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
{
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
if (strcmp(master_row[0], global_system_variables.collation_server->name))
{
errmsg= "The slave I/O thread stops because master and slave have \
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
different values for the COLLATION_SERVER global variable. The values must \
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
be equal for the Statement-format replication to work";
err_code= ER_SLAVE_FATAL_ERROR;
sprintf(err_buff, ER(err_code), errmsg);
goto err;
}
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
}
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
else if (is_network_error(mysql_errno(mysql)))
{
mi->report(WARNING_LEVEL, mysql_errno(mysql),
"Get master COLLATION_SERVER failed with error: %s", mysql_error(mysql));
goto network_err;
}
else if (mysql_errno(mysql) != ER_UNKNOWN_SYSTEM_VARIABLE)
{
/* Fatal error */
errmsg= "The slave I/O thread stops because a fatal error is encountered \
when it try to get the value of COLLATION_SERVER global variable from master.";
err_code= mysql_errno(mysql);
sprintf(err_buff, "%s Error: %s", errmsg, mysql_error(mysql));
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
goto err;
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
}
else
mi->report(WARNING_LEVEL, ER_UNKNOWN_SYSTEM_VARIABLE,
"Unknown system variable 'COLLATION_SERVER' on master, \
maybe it is a *VERY OLD MASTER*. *NOTE*: slave may experience \
inconsistency if replicated data deals with collation.");
if (master_res)
{
mysql_free_result(master_res);
master_res= NULL;
}
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
}
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
WL#1264 "Per-thread time zone support infrastructure". Added basic per-thread time zone functionality (based on public domain elsie-code). Now user can select current time zone (from the list of time zones described in system tables). All NOW-like functions honor this time zone, values of TIMESTAMP type are interpreted as values in this time zone, so now our TIMESTAMP type behaves similar to Oracle's TIMESTAMP WITH LOCAL TIME ZONE (or proper PostgresSQL type). WL#1266 "CONVERT_TZ() - basic time with time zone conversion function". Fixed problems described in Bug #2336 (Different number of warnings when inserting bad datetime as string or as number). This required reworking of datetime realted warning hadling (they now generated at Field object level not in conversion functions). Optimization: Now Field class descendants use table->in_use member instead of current_thd macro.
2004-06-18 08:11:31 +02:00
/*
Perform analogous check for time zone. Theoretically we also should
perform check here to verify that SYSTEM time zones are the same on
slave and master, but we can't rely on value of @@system_time_zone
variable (it is time zone abbreviation) since it determined at start
time and so could differ for slave and master even if they are really
in the same system time zone. So we are omiting this check and just
relying on documentation. Also according to Monty there are many users
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
who are using replication between servers in various time zones. Hence
such check will broke everything for them. (And now everything will
work for them because by default both their master and slave will have
WL#1264 "Per-thread time zone support infrastructure". Added basic per-thread time zone functionality (based on public domain elsie-code). Now user can select current time zone (from the list of time zones described in system tables). All NOW-like functions honor this time zone, values of TIMESTAMP type are interpreted as values in this time zone, so now our TIMESTAMP type behaves similar to Oracle's TIMESTAMP WITH LOCAL TIME ZONE (or proper PostgresSQL type). WL#1266 "CONVERT_TZ() - basic time with time zone conversion function". Fixed problems described in Bug #2336 (Different number of warnings when inserting bad datetime as string or as number). This required reworking of datetime realted warning hadling (they now generated at Field object level not in conversion functions). Optimization: Now Field class descendants use table->in_use member instead of current_thd macro.
2004-06-18 08:11:31 +02:00
'SYSTEM' time zone).
Last part of WL#1062: better replication of timezones: no more use of SET ONE_SHOT; storing tz info directly in event (if this info is needed), it's now allowed to have different global tz on master and slave.
2005-03-22 00:26:12 +01:00
This check is only necessary for 4.x masters (and < 5.0.4 masters but
those were alpha).
WL#1264 "Per-thread time zone support infrastructure". Added basic per-thread time zone functionality (based on public domain elsie-code). Now user can select current time zone (from the list of time zones described in system tables). All NOW-like functions honor this time zone, values of TIMESTAMP type are interpreted as values in this time zone, so now our TIMESTAMP type behaves similar to Oracle's TIMESTAMP WITH LOCAL TIME ZONE (or proper PostgresSQL type). WL#1266 "CONVERT_TZ() - basic time with time zone conversion function". Fixed problems described in Bug #2336 (Different number of warnings when inserting bad datetime as string or as number). This required reworking of datetime realted warning hadling (they now generated at Field object level not in conversion functions). Optimization: Now Field class descendants use table->in_use member instead of current_thd macro.
2004-06-18 08:11:31 +02:00
*/
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
if (*mysql->server_version == '4')
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
{
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
master_res= NULL;
if (!mysql_real_query(mysql, STRING_WITH_LEN("SELECT @@GLOBAL.TIME_ZONE")) &&
(master_res= mysql_store_result(mysql)) &&
(master_row= mysql_fetch_row(master_res)))
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
{
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
if (strcmp(master_row[0],
global_system_variables.time_zone->get_name()->ptr()))
{
errmsg= "The slave I/O thread stops because master and slave have \
WL#1264 "Per-thread time zone support infrastructure". Added basic per-thread time zone functionality (based on public domain elsie-code). Now user can select current time zone (from the list of time zones described in system tables). All NOW-like functions honor this time zone, values of TIMESTAMP type are interpreted as values in this time zone, so now our TIMESTAMP type behaves similar to Oracle's TIMESTAMP WITH LOCAL TIME ZONE (or proper PostgresSQL type). WL#1266 "CONVERT_TZ() - basic time with time zone conversion function". Fixed problems described in Bug #2336 (Different number of warnings when inserting bad datetime as string or as number). This required reworking of datetime realted warning hadling (they now generated at Field object level not in conversion functions). Optimization: Now Field class descendants use table->in_use member instead of current_thd macro.
2004-06-18 08:11:31 +02:00
different values for the TIME_ZONE global variable. The values must \
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
be equal for the Statement-format replication to work";
err_code= ER_SLAVE_FATAL_ERROR;
sprintf(err_buff, ER(err_code), errmsg);
goto err;
}
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
}
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
else if (is_network_error(mysql_errno(mysql)))
{
mi->report(WARNING_LEVEL, mysql_errno(mysql),
"Get master TIME_ZONE failed with error: %s", mysql_error(mysql));
goto network_err;
}
else
{
/* Fatal error */
errmsg= "The slave I/O thread stops because a fatal error is encountered \
when it try to get the value of TIME_ZONE global variable from master.";
err_code= mysql_errno(mysql);
sprintf(err_buff, "%s Error: %s", errmsg, mysql_error(mysql));
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
goto err;
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
}
if (master_res)
{
mysql_free_result(master_res);
master_res= NULL;
}
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
}
Fix for BUG#7965 "Slave_IO_State Stuck at 'Checking Master Version'": Working around hang of master < 3.23.50 on SELECT @@unknown_var (to enable 3.23.49->4.1.10 replication)
2005-01-17 21:26:14 +01:00
err:
work to enable reading 3.23 logs - not yet finished moved fail-safe replication routines from sql_repl.cc to repl_failsafe.cc write start event only in the first log
2001-11-11 06:24:12 +01:00
if (errmsg)
{
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
if (master_res)
mysql_free_result(master_res);
Bug #31316 Report server id clashes in SHOW SLAVE STATUS "Server_IO_State" field Critical error messages from get_master_version_and_clock() were written only to the slave errorlog while Show slave status did not display any incident happened. Although the artifact was reported for a particular --replicate-same-server-id related issue the fix refines all critical error reporting with deploying rli->report(). The test for the bug covers only --replicate-same-server-id error reporting.
2008-02-20 22:18:01 +01:00
DBUG_ASSERT(err_code != 0);
Bug #43414 Parenthesis (and other) warnings compiling MySQL with gcc 4.3.2 Cleaning up warnings not present in 5.0.
2009-09-23 15:21:29 +02:00
mi->report(ERROR_LEVEL, err_code, "%s", err_buff);
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
work to enable reading 3.23 logs - not yet finished moved fail-safe replication routines from sql_repl.cc to repl_failsafe.cc write start event only in the first log
2001-11-11 06:24:12 +01:00
}
Implementation of WL#1824 "Add replication of character set variables in 4.1", by binlogging some SET ONE_SHOT CHARACTER_SETetc, which will be enough until we have it more compact and more complete in 5.0. With the present patch, replication will work ok between 4.1.3 master and slaves, as long as: - master and slave have the same GLOBAL.COLLATION_SERVER - COLLATION_DATABASE and CHARACTER_SET_DATABASE are not used - application does not use the fact that table is created with charset of the USEd db (BUG#2326). all of which are not too hard to fulfill. ONE_SHOT is reserved for internal use of mysqlbinlog|mysql and works only for charsets, so we give error if used for non-charset vars. Fix for BUG#3875 "mysqlbinlog produces wrong ouput if query uses variables containing quotes" and BUG#3943 "Queries with non-ASCII literals are not replicated properly after SET NAMES". Detecting that master and slave have different global charsets or server ids.
2004-06-03 23:17:18 +02:00
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
network_err:
if (master_res)
mysql_free_result(master_res);
DBUG_RETURN(2);
work to enable reading 3.23 logs - not yet finished moved fail-safe replication routines from sql_repl.cc to repl_failsafe.cc write start event only in the first log
2001-11-11 06:24:12 +01:00
}
* Fix for BUG#1248: "LOAD DATA FROM MASTER drops the slave's db unexpectedly". Now LOAD DATA FROM MASTER does not drop the database, instead it only tries to create it, and drops/creates table-by-table. * replicate_wild_ignore_table='db1.%' is now considered as "ignore the 'db1' database as a whole", as it already works for CREATE DATABASE and DROP DATABASE.
2003-09-11 23:17:28 +02:00
/*
Used by fetch_master_table (used by LOAD TABLE tblname FROM MASTER and LOAD
DATA FROM MASTER). Drops the table (if 'overwrite' is true) and recreates it
from the dump. Honours replication inclusion/exclusion rules.
when we update thd->db in replication, it's safer to update thd->db_length too. This does not fix any known bug, but is still a good idea.
2004-09-07 14:57:54 +02:00
db must be non-zero (guarded by assertion).
* Fix for BUG#1248: "LOAD DATA FROM MASTER drops the slave's db unexpectedly". Now LOAD DATA FROM MASTER does not drop the database, instead it only tries to create it, and drops/creates table-by-table. * replicate_wild_ignore_table='db1.%' is now considered as "ignore the 'db1' database as a whole", as it already works for CREATE DATABASE and DROP DATABASE.
2003-09-11 23:17:28 +02:00
RETURN VALUES
0 success
1 error
*/
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
static int create_table_from_dump(THD* thd, MYSQL *mysql, const char* db,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
const char* table_name, bool overwrite)
Import changeset
2000-07-31 21:29:14 +02:00
{
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
ulong packet_len;
when we update thd->db in replication, it's safer to update thd->db_length too. This does not fix any known bug, but is still a good idea.
2004-09-07 14:57:54 +02:00
char *query, *save_db;
uint32 save_db_length;
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
Vio* save_vio;
HA_CHECK_OPT check_opt;
Import changeset
2000-07-31 21:29:14 +02:00
TABLE_LIST tables;
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
int error= 1;
handler *file;
Fixed compiler warnings (Mostly VC++): - Removed not used variables - Changed some ulong parameters/variables to ulonglong (possible serious bug) - Added casts to get rid of safe assignment from longlong to long (and similar) - Added casts to function parameters - Fixed signed/unsigned compares - Added some constructores to structures - Removed some not portable constructs Better fix for bug Bug #21428 "skipped 9 bytes from file: socket (3)" on "mysqladmin shutdown" (Added new parameter to net_clear() to define when we want the communication buffer to be emptied)
2006-11-30 02:40:42 +01:00
ulonglong save_options;
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
NET *net= &mysql->net;
manual merge 5.0-runtime -> 5.1->runtime, with 25411 part I
2007-04-26 05:38:12 +02:00
const char *found_semicolon= NULL;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("create_table_from_dump");
Fixes after merge
2003-10-08 11:01:58 +02:00
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
packet_len= my_net_read(net); // read create table statement
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
if (packet_len == packet_error)
{
post-review fixes
2004-11-12 13:34:00 +01:00
my_message(ER_MASTER_NET_READ, ER(ER_MASTER_NET_READ), MYF(0));
Fixes after merge
2003-10-08 11:01:58 +02:00
DBUG_RETURN(1);
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
}
if (net->read_pos[0] == 255) // error from master
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
char *err_msg;
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
err_msg= (char*) net->read_pos + ((mysql->server_capabilities &
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
CLIENT_PROTOCOL_41) ?
3+SQLSTATE_LENGTH+1 : 3);
now my_printf_error is not better then my_error, but my_error call is shorter used only one implementation of format parser of (printf) fixed multistatement
2004-11-13 18:35:51 +01:00
my_error(ER_MASTER, MYF(0), err_msg);
Fixes after merge
2003-10-08 11:01:58 +02:00
DBUG_RETURN(1);
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
}
Import changeset
2000-07-31 21:29:14 +02:00
thd->command = COM_TABLE_DUMP;
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
if (!(query = thd->strmake((char*) net->read_pos, packet_len)))
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
{
sql_print_error("create_table_from_dump: out of memory");
post-review fixes
2004-11-12 13:34:00 +01:00
my_message(ER_GET_ERRNO, "Out of memory", MYF(0));
Fixes after merge
2003-10-08 11:01:58 +02:00
DBUG_RETURN(1);
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
}
Bug #38816: kill + flush tables with read lock + stored procedures causes crashes! The problem of that bugreport was mostly fixed by the patch for bug 38691. However, attached test case focused on another crash or valgrind warning problem: SHOW PROCESSLIST query accesses freed memory of SP instruction that run in a parallel connection. Changes of thd->query/thd->query_length in dangerous places have been guarded with the per-thread LOCK_thd_data mutex (the THD::LOCK_delete mutex has been renamed to THD::LOCK_thd_data).
2009-07-24 17:58:58 +02:00
thd->set_query(query, packet_len);
Rename: query_error -> is_slave_error. Add comments.
2007-10-19 23:20:38 +02:00
thd->is_slave_error = 0;
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
Fixes after merge
2003-10-08 11:01:58 +02:00
bzero((char*) &tables,sizeof(tables));
tables.db = (char*)db;
First stage of table definition cache Split TABLE to TABLE and TABLE_SHARE (TABLE_SHARE is still allocated as part of table, will be fixed soon) Created Field::make_field() and made Field_num::make_field() to call this Added 'TABLE_SHARE->db' that points to database name; Changed all usage of table_cache_key as database name to use this instead Changed field->table_name to point to pointer to alias. This allows us to change alias for a table by just updating one pointer. Renamed TABLE_SHARE->real_name to table_name Renamed TABLE->table_name to alias Renamed TABLE_LIST->real_name to table_name
2005-01-06 12:00:13 +01:00
tables.alias= tables.table_name= (char*)table_name;
after review changes
2004-04-07 12:25:24 +02:00
Fixes after merge
2003-10-08 11:01:58 +02:00
/* Drop the table if 'overwrite' is true */
Bug#12713 "Error in a stored function called from a SELECT doesn't cause ROLLBACK of statement", part 1. Review fixes. Do not send OK/EOF packets to the client until we reached the end of the current statement. This is a consolidation, to keep the functionality that is shared by all SQL statements in one place in the server. Currently this functionality includes: - close_thread_tables() - log_slow_statement(). After this patch and the subsequent patch for Bug#12713, it shall also include: - ha_autocommit_or_rollback() - net_end_statement() - query_cache_end_of_result(). In future it may also include: - mysql_reset_thd_for_next_command().
2007-12-12 16:21:01 +01:00
if (overwrite)
Fixes after merge
2003-10-08 11:01:58 +02:00
{
Bug#12713 "Error in a stored function called from a SELECT doesn't cause ROLLBACK of statement", part 1. Review fixes. Do not send OK/EOF packets to the client until we reached the end of the current statement. This is a consolidation, to keep the functionality that is shared by all SQL statements in one place in the server. Currently this functionality includes: - close_thread_tables() - log_slow_statement(). After this patch and the subsequent patch for Bug#12713, it shall also include: - ha_autocommit_or_rollback() - net_end_statement() - query_cache_end_of_result(). In future it may also include: - mysql_reset_thd_for_next_command().
2007-12-12 16:21:01 +01:00
if (mysql_rm_table(thd,&tables,1,0)) /* drop if exists */
{
sql_print_error("create_table_from_dump: failed to drop the table");
goto err;
}
else
{
/* Clear the OK result of mysql_rm_table(). */
thd->main_da.reset_diagnostics_area();
}
Fixes after merge
2003-10-08 11:01:58 +02:00
}
* Fix for BUG#1248: "LOAD DATA FROM MASTER drops the slave's db unexpectedly". Now LOAD DATA FROM MASTER does not drop the database, instead it only tries to create it, and drops/creates table-by-table. * replicate_wild_ignore_table='db1.%' is now considered as "ignore the 'db1' database as a whole", as it already works for CREATE DATABASE and DROP DATABASE.
2003-09-11 23:17:28 +02:00
/* Create the table. We do not want to log the "create table" statement */
use env vars for socket/port in connect in mysqltest --local-master option in mysql-test-run do not log CREATE TABLE when doing table dump fix replication to slave off 3.23 master
2001-11-18 01:23:46 +01:00
save_options = thd->options;
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd->options &= ~ (OPTION_BIN_LOG);
thd_proc_info(thd, "Creating table from master dump");
keep bk happy
2001-01-22 03:46:32 +01:00
// save old db in case we are creating in a different database
* Fix for BUG#1248: "LOAD DATA FROM MASTER drops the slave's db unexpectedly". Now LOAD DATA FROM MASTER does not drop the database, instead it only tries to create it, and drops/creates table-by-table. * replicate_wild_ignore_table='db1.%' is now considered as "ignore the 'db1' database as a whole", as it already works for CREATE DATABASE and DROP DATABASE.
2003-09-11 23:17:28 +02:00
save_db = thd->db;
when we update thd->db in replication, it's safer to update thd->db_length too. This does not fix any known bug, but is still a good idea.
2004-09-07 14:57:54 +02:00
save_db_length= thd->db_length;
updates for LOAD DATA FROM MASTER + some cleanup of replication code
2001-05-29 03:18:23 +02:00
thd->db = (char*)db;
Remove compiler warnings and remove not used variables (Found during build process)
2005-02-25 15:53:22 +01:00
DBUG_ASSERT(thd->db != 0);
when we update thd->db in replication, it's safer to update thd->db_length too. This does not fix any known bug, but is still a good idea.
2004-09-07 14:57:54 +02:00
thd->db_length= strlen(thd->db);
manual merge 5.0-runtime -> 5.1->runtime, with 25411 part I
2007-04-26 05:38:12 +02:00
mysql_parse(thd, thd->query, packet_len, &found_semicolon); // run create table
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
thd->db = save_db; // leave things the way the were before
when we update thd->db in replication, it's safer to update thd->db_length too. This does not fix any known bug, but is still a good idea.
2004-09-07 14:57:54 +02:00
thd->db_length= save_db_length;
use env vars for socket/port in connect in mysqltest --local-master option in mysql-test-run do not log CREATE TABLE when doing table dump fix replication to slave off 3.23 master
2001-11-18 01:23:46 +01:00
thd->options = save_options;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Rename: query_error -> is_slave_error. Add comments.
2007-10-19 23:20:38 +02:00
if (thd->is_slave_error)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err; // mysql_parse took care of the error send
Import changeset
2000-07-31 21:29:14 +02:00
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Opening master dump table");
Bug#12713 "Error in a stored function called from a SELECT doesn't cause ROLLBACK of statement", part 1. Review fixes. Do not send OK/EOF packets to the client until we reached the end of the current statement. This is a consolidation, to keep the functionality that is shared by all SQL statements in one place in the server. Currently this functionality includes: - close_thread_tables() - log_slow_statement(). After this patch and the subsequent patch for Bug#12713, it shall also include: - ha_autocommit_or_rollback() - net_end_statement() - query_cache_end_of_result(). In future it may also include: - mysql_reset_thd_for_next_command().
2007-12-12 16:21:01 +01:00
thd->main_da.reset_diagnostics_area(); /* cleanup from CREATE_TABLE */
Bug#26379 - Combination of FLUSH TABLE and REPAIR TABLE corrupts a MERGE table Bug 26867 - LOCK TABLES + REPAIR + merge table result in memory/cpu hogging Bug 26377 - Deadlock with MERGE and FLUSH TABLE Bug 25038 - Waiting TRUNCATE Bug 25700 - merge base tables get corrupted by optimize/analyze/repair table Bug 30275 - Merge tables: flush tables or unlock tables causes server to crash Bug 19627 - temporary merge table locking Bug 27660 - Falcon: merge table possible Bug 30273 - merge tables: Can't lock file (errno: 155) The problems were: Bug 26379 - Combination of FLUSH TABLE and REPAIR TABLE corrupts a MERGE table 1. A thread trying to lock a MERGE table performs busy waiting while REPAIR TABLE or a similar table administration task is ongoing on one or more of its MyISAM tables. 2. A thread trying to lock a MERGE table performs busy waiting until all threads that did REPAIR TABLE or similar table administration tasks on one or more of its MyISAM tables in LOCK TABLES segments do UNLOCK TABLES. The difference against problem #1 is that the busy waiting takes place *after* the administration task. It is terminated by UNLOCK TABLES only. 3. Two FLUSH TABLES within a LOCK TABLES segment can invalidate the lock. This does *not* require a MERGE table. The first FLUSH TABLES can be replaced by any statement that requires other threads to reopen the table. In 5.0 and 5.1 a single FLUSH TABLES can provoke the problem. Bug 26867 - LOCK TABLES + REPAIR + merge table result in memory/cpu hogging Trying DML on a MERGE table, which has a child locked and repaired by another thread, made an infinite loop in the server. Bug 26377 - Deadlock with MERGE and FLUSH TABLE Locking a MERGE table and its children in parent-child order and flushing the child deadlocked the server. Bug 25038 - Waiting TRUNCATE Truncating a MERGE child, while the MERGE table was in use, let the truncate fail instead of waiting for the table to become free. Bug 25700 - merge base tables get corrupted by optimize/analyze/repair table Repairing a child of an open MERGE table corrupted the child. It was necessary to FLUSH the child first. Bug 30275 - Merge tables: flush tables or unlock tables causes server to crash Flushing and optimizing locked MERGE children crashed the server. Bug 19627 - temporary merge table locking Use of a temporary MERGE table with non-temporary children could corrupt the children. Temporary tables are never locked. So we do now prohibit non-temporary chidlren of a temporary MERGE table. Bug 27660 - Falcon: merge table possible It was possible to create a MERGE table with non-MyISAM children. Bug 30273 - merge tables: Can't lock file (errno: 155) This was a Windows-only bug. Table administration statements sometimes failed with "Can't lock file (errno: 155)". These bugs are fixed by a new implementation of MERGE table open. When opening a MERGE table in open_tables() we do now add the child tables to the list of tables to be opened by open_tables() (the "query_list"). The children are not opened in the handler at this stage. After opening the parent, open_tables() opens each child from the now extended query_list. When the last child is opened, we remove the children from the query_list again and attach the children to the parent. This behaves similar to the old open. However it does not open the MyISAM tables directly, but grabs them from the already open children. When closing a MERGE table in close_thread_table() we detach the children only. Closing of the children is done implicitly because they are in thd->open_tables. For more detail see the comment at the top of ha_myisammrg.cc. Changed from open_ltable() to open_and_lock_tables() in all places that can be relevant for MERGE tables. The latter can handle tables added to the list on the fly. When open_ltable() was used in a loop over a list of tables, the list must be temporarily terminated after every table for open_and_lock_tables(). table_list->required_type is set to FRMTYPE_TABLE to avoid open of special tables. Handling of derived tables is suppressed. These details are handled by the new function open_n_lock_single_table(), which has nearly the same signature as open_ltable() and can replace it in most cases. In reopen_tables() some of the tables open by a thread can be closed and reopened. When a MERGE child is affected, the parent must be closed and reopened too. Closing of the parent is forced before the first child is closed. Reopen happens in the order of thd->open_tables. MERGE parents do not attach their children automatically at open. This is done after all tables are reopened. So all children are open when attaching them. Special lock handling like mysql_lock_abort() or mysql_lock_remove() needs to be suppressed for MERGE children or forwarded to the parent. This depends on the situation. In loops over all open tables one suppresses child lock handling. When a single table is touched, forwarding is done. Behavioral changes: =================== This patch changes the behavior of temporary MERGE tables. Temporary MERGE must have temporary children. The old behavior was wrong. A temporary table is not locked. Hence even non-temporary children were not locked. See Bug 19627 - temporary merge table locking. You cannot change the union list of a non-temporary MERGE table when LOCK TABLES is in effect. The following does *not* work: CREATE TABLE m1 ... ENGINE=MRG_MYISAM ...; LOCK TABLES t1 WRITE, t2 WRITE, m1 WRITE; ALTER TABLE m1 ... UNION=(t1,t2) ...; However, you can do this with a temporary MERGE table. You cannot create a MERGE table with CREATE ... SELECT, neither as a temporary MERGE table, nor as a non-temporary MERGE table. CREATE TABLE m1 ... ENGINE=MRG_MYISAM ... SELECT ...; Gives error message: table is not BASE TABLE.
2007-11-15 20:25:43 +01:00
/*
Note: If this function starts to fail for MERGE tables,
change the next two lines to these:
tables.table= NULL; // was set by mysql_rm_table()
if (!open_n_lock_single_table(thd, &tables, TL_WRITE))
*/
* Fix for BUG#1248: "LOAD DATA FROM MASTER drops the slave's db unexpectedly". Now LOAD DATA FROM MASTER does not drop the database, instead it only tries to create it, and drops/creates table-by-table. * replicate_wild_ignore_table='db1.%' is now considered as "ignore the 'db1' database as a whole", as it already works for CREATE DATABASE and DROP DATABASE.
2003-09-11 23:17:28 +02:00
tables.lock_type = TL_WRITE;
WL#3984 (Revise locking of mysql.general_log and mysql.slow_log) Bug#25422 (Hang with log tables) Bug 17876 (Truncating mysql.slow_log in a SP after using cursor locks the thread) Bug 23044 (Warnings on flush of a log table) Bug 29129 (Resetting general_log while the GLOBAL READ LOCK is set causes a deadlock) Prior to this fix, the server would hang when performing concurrent ALTER TABLE or TRUNCATE TABLE statements against the LOG TABLES, which are mysql.general_log and mysql.slow_log. The root cause traces to the following code: in sql_base.cc, open_table() if (table->in_use != thd) { /* wait_for_condition will unlock LOCK_open for us */ wait_for_condition(thd, &LOCK_open, &COND_refresh); } The problem with this code is that the current implementation of the LOGGER creates 'fake' THD objects, like - Log_to_csv_event_handler::general_log_thd - Log_to_csv_event_handler::slow_log_thd which are not associated to a real thread running in the server, so that waiting for these non-existing threads to release table locks cause the dead lock. In general, the design of Log_to_csv_event_handler does not fit into the general architecture of the server, so that the concept of general_log_thd and slow_log_thd has to be abandoned: - this implementation does not work with table locking - it will not work with commands like SHOW PROCESSLIST - having the log tables always opened does not integrate well with DDL operations / FLUSH TABLES / SET GLOBAL READ_ONLY With this patch, the fundamental design of the LOGGER has been changed to: - always open and close a log table when writing a log - remove totally the usage of fake THD objects - clarify how locking of log tables is implemented in general. See WL#3984 for details related to the new locking design. Additional changes (misc bugs exposed and fixed): 1) mysqldump which would ignore some tables in dump_all_tables_in_db(), but forget to ignore the same in dump_all_views_in_db(). 2) mysqldump would also issue an empty "LOCK TABLE" command when all the tables to lock are to be ignored (numrows == 0), instead of not issuing the query. 3) Internal errors handlers could intercept errors but not warnings (see sql_error.cc). 4) Implementing a nested call to open tables, for the performance schema tables, exposed an existing bug in remove_table_from_cache(), which would perform: in_use->some_tables_deleted=1; against another thread, without any consideration about thread locking. This call inside remove_table_from_cache() was not required anyway, since calling mysql_lock_abort() takes care of aborting -- cleanly -- threads that might hold a lock on a table. This line (in_use->some_tables_deleted=1) has been removed.
2007-07-27 08:31:06 +02:00
if (!open_ltable(thd, &tables, TL_WRITE, 0))
keep bk happy
2001-01-22 03:46:32 +01:00
{
sql_print_error("create_table_from_dump: could not open created table");
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
goto err;
keep bk happy
2001-01-22 03:46:32 +01:00
}
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
file = tables.table->file;
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Reading master dump table data");
* Fix for BUG#1248: "LOAD DATA FROM MASTER drops the slave's db unexpectedly". Now LOAD DATA FROM MASTER does not drop the database, instead it only tries to create it, and drops/creates table-by-table. * replicate_wild_ignore_table='db1.%' is now considered as "ignore the 'db1' database as a whole", as it already works for CREATE DATABASE and DROP DATABASE.
2003-09-11 23:17:28 +02:00
/* Copy the data file */
keep bk happy
2001-01-22 03:46:32 +01:00
if (file->net_read_dump(net))
{
post-review fixes
2004-11-12 13:34:00 +01:00
my_message(ER_MASTER_NET_READ, ER(ER_MASTER_NET_READ), MYF(0));
* Fix for BUG#1248: "LOAD DATA FROM MASTER drops the slave's db unexpectedly". Now LOAD DATA FROM MASTER does not drop the database, instead it only tries to create it, and drops/creates table-by-table. * replicate_wild_ignore_table='db1.%' is now considered as "ignore the 'db1' database as a whole", as it already works for CREATE DATABASE and DROP DATABASE.
2003-09-11 23:17:28 +02:00
sql_print_error("create_table_from_dump: failed in\
Import changeset
2000-07-31 21:29:14 +02:00
handler::net_read_dump()");
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
goto err;
keep bk happy
2001-01-22 03:46:32 +01:00
}
Import changeset
2000-07-31 21:29:14 +02:00
check_opt.init();
mi_check flags cleanup CHECK/REPAIR syntax cleanup REPAIR ... USE_FRM syntax
2002-03-13 18:20:17 +01:00
check_opt.flags|= T_VERY_SILENT | T_CALC_CHECKSUM | T_QUICK;
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Rebuilding the index on master dump table");
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
We do not want repair() to spam us with messages
just send them to the error log, and report the failure in case of
problems.
*/
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
save_vio = thd->net.vio;
keep bk happy
2001-01-22 03:46:32 +01:00
thd->net.vio = 0;
* Fix for BUG#1248: "LOAD DATA FROM MASTER drops the slave's db unexpectedly". Now LOAD DATA FROM MASTER does not drop the database, instead it only tries to create it, and drops/creates table-by-table. * replicate_wild_ignore_table='db1.%' is now considered as "ignore the 'db1' database as a whole", as it already works for CREATE DATABASE and DROP DATABASE.
2003-09-11 23:17:28 +02:00
/* Rebuild the index file from the copied data file (with REPAIR) */
WL#2645 (CHECK TABLE FOR UPGRADE) necessary implementation in the server mysql_upgrade script added
2006-02-17 07:52:32 +01:00
error=file->ha_repair(thd,&check_opt) != 0;
Import changeset
2000-07-31 21:29:14 +02:00
thd->net.vio = save_vio;
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
if (error)
Table definition cache, part 2 The table opening process now works the following way: - Create common TABLE_SHARE object - Read the .frm file and unpack it into the TABLE_SHARE object - Create a TABLE object based on the information in the TABLE_SHARE object and open a handler to the table object Other noteworthy changes: - In TABLE_SHARE the most common strings are now LEX_STRING's - Better error message when table is not found - Variable table_cache is now renamed 'table_open_cache' - New variable 'table_definition_cache' that is the number of table defintions that will be cached - strxnmov() calls are now fixed to avoid overflows - strxnmov() will now always add one end \0 to result - engine objects are now created with a TABLE_SHARE object instead of a TABLE object. - After creating a field object one must call field->init(table) before using it - For a busy system this change will give you: - Less memory usage for table object - Faster opening of tables (if it's has been in use or is in table definition cache) - Allow you to cache many table definitions objects - Faster drop of table
2005-11-23 21:45:02 +01:00
my_error(ER_INDEX_REBUILD, MYF(0), tables.table->s->table_name.str);
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
err:
Import changeset
2000-07-31 21:29:14 +02:00
close_thread_tables(thd);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
DBUG_RETURN(error);
Import changeset
2000-07-31 21:29:14 +02:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
int fetch_master_table(THD *thd, const char *db_name, const char *table_name,
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
Master_info *mi, MYSQL *mysql, bool overwrite)
Import changeset
2000-07-31 21:29:14 +02:00
{
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
int error= 1;
const char *errmsg=0;
bool called_connected= (mysql != NULL);
DBUG_ENTER("fetch_master_table");
DBUG_PRINT("enter", ("db_name: '%s' table_name: '%s'",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
db_name,table_name));
Import changeset
2000-07-31 21:29:14 +02:00
merge
2001-06-01 23:51:23 +02:00
if (!called_connected)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
{
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
if (!(mysql = mysql_init(NULL)))
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
{
DBUG_RETURN(1);
}
merge
2001-06-01 23:51:23 +02:00
if (connect_to_master(thd, mysql, mi))
updates for LOAD DATA FROM MASTER + some cleanup of replication code
2001-05-29 03:18:23 +02:00
{
now my_printf_error is not better then my_error, but my_error call is shorter used only one implementation of format parser of (printf) fixed multistatement
2004-11-13 18:35:51 +01:00
my_error(ER_CONNECT_TO_MASTER, MYF(0), mysql_error(mysql));
Bug#19437 (Connection refused by server: "2002 Can't connect... /master.sock"): Clearing active VIO before calling mysql_close() in the slave I/O thread.
2006-06-20 20:46:45 +02:00
/*
We need to clear the active VIO since, theoretically, somebody
might issue an awake() on this thread. If we are then in the
middle of closing and destroying the VIO inside the
mysql_close(), we will have a problem.
*/
#ifdef SIGNAL_WITH_VIO_CLOSE
thd->clear_active_vio();
#endif
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
mysql_close(mysql);
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
DBUG_RETURN(1);
updates for LOAD DATA FROM MASTER + some cleanup of replication code
2001-05-29 03:18:23 +02:00
}
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
if (thd->killed)
goto err;
updates for LOAD DATA FROM MASTER + some cleanup of replication code
2001-05-29 03:18:23 +02:00
}
Import changeset
2000-07-31 21:29:14 +02:00
sql/slave.cc fixed bad merge
2001-06-02 01:53:52 +02:00
if (request_table_dump(mysql, db_name, table_name))
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
{
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
error= ER_UNKNOWN_ERROR;
errmsg= "Failed on table dump request";
Fixed bug that caused client to hang because mysqld never did send an error message if the table open or the index creation failed. Updated portuguese error messages. Fix for OS/2 that affected CHECK TABLE.
2001-06-01 23:03:16 +02:00
goto err;
}
Merge with 4.0.16
2003-10-07 14:42:26 +02:00
if (create_table_from_dump(thd, mysql, db_name,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
table_name, overwrite))
Merge with 4.0.16
2003-10-07 14:42:26 +02:00
goto err; // create_table_from_dump have sent the error already
Import changeset
2000-07-31 21:29:14 +02:00
error = 0;
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
Import changeset
2000-07-31 21:29:14 +02:00
err:
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
if (!called_connected)
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
mysql_close(mysql);
Proposed fix for bug #3412 (embedded server: prepared statement returns empty recordset where some records should be found)
2004-05-28 12:59:29 +02:00
if (errmsg && thd->vio_ok())
errors without code removed net_printf/send_error calls replaced by my_error family functions -1/1 (sent/unsent) error reporting removed (WL#2133)
2004-10-20 03:04:37 +02:00
my_message(error, errmsg, MYF(0));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
DBUG_RETURN(test(error)); // Return 1 on error
Import changeset
2000-07-31 21:29:14 +02:00
}
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
open binlog index file *before* tc_log->open() and binlog itself *after*
2005-02-01 15:36:48 +01:00
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
static bool wait_for_relay_log_space(Relay_log_info* rli)
relay_log_space_limit DBUG_ macro cleanup buffer boundary cleanup This changeset, although not fully tested, works for me better than anything I've had so far, including what is in the repository. I will push it unless something crashes while I am writing this :-)
2002-04-02 06:46:23 +02:00
{
Fixed bug in wait_for_relay_log_space() Added checking of sem_init() in configure.
2002-06-02 16:04:16 +02:00
bool slave_killed=0;
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
Master_info* mi = rli->mi;
Merge with 4.0.14
2003-08-11 21:44:43 +02:00
const char *save_proc_info;
relay_log_space_limit DBUG_ macro cleanup buffer boundary cleanup This changeset, although not fully tested, works for me better than anything I've had so far, including what is in the repository. I will push it unless something crashes while I am writing this :-)
2002-04-02 06:46:23 +02:00
THD* thd = mi->io_thd;
DBUG_ENTER("wait_for_relay_log_space");
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
relay_log_space_limit DBUG_ macro cleanup buffer boundary cleanup This changeset, although not fully tested, works for me better than anything I've had so far, including what is in the repository. I will push it unless something crashes while I am writing this :-)
2002-04-02 06:46:23 +02:00
pthread_mutex_lock(&rli->log_space_lock);
Merge with 4.0.14
2003-08-11 21:44:43 +02:00
save_proc_info= thd->enter_cond(&rli->log_space_cond,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
&rli->log_space_lock,
"\
Change word order of a few PROCESSLIST messages to correspond to order used in START SLAVE syntax.
2003-11-21 19:35:33 +01:00
Waiting for the slave SQL thread to free enough relay log space");
relay_log_space_limit DBUG_ macro cleanup buffer boundary cleanup This changeset, although not fully tested, works for me better than anything I've had so far, including what is in the repository. I will push it unless something crashes while I am writing this :-)
2002-04-02 06:46:23 +02:00
while (rli->log_space_limit < rli->log_space_total &&
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
!(slave_killed=io_slave_killed(thd,mi)) &&
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
!rli->ignore_log_space_limit)
relay_log_space_limit DBUG_ macro cleanup buffer boundary cleanup This changeset, although not fully tested, works for me better than anything I've had so far, including what is in the repository. I will push it unless something crashes while I am writing this :-)
2002-04-02 06:46:23 +02:00
pthread_cond_wait(&rli->log_space_cond, &rli->log_space_lock);
Simplified a test. thd->enter_cond() and exit_cond(), so that the I/O thread accepts to stop when it's waiting for relay log space. Reset ignore_log_space_limit to 0 when the SQL thread terminates.
2003-06-15 12:01:51 +02:00
thd->exit_cond(save_proc_info);
relay_log_space_limit DBUG_ macro cleanup buffer boundary cleanup This changeset, although not fully tested, works for me better than anything I've had so far, including what is in the repository. I will push it unless something crashes while I am writing this :-)
2002-04-02 06:46:23 +02:00
DBUG_RETURN(slave_killed);
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
open binlog index file *before* tc_log->open() and binlog itself *after*
2005-02-01 15:36:48 +01:00
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
/*
Builds a Rotate from the ignored events' info and writes it to relay log.
open binlog index file *before* tc_log->open() and binlog itself *after*
2005-02-01 15:36:48 +01:00
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
SYNOPSIS
write_ignored_events_info_to_relay_log()
thd pointer to I/O thread's thd
mi
DESCRIPTION
Slave I/O thread, going to die, must leave a durable trace of the
ignored events' end position for the use of the slave SQL thread, by
calling this function. Only that thread can call it (see assertion).
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static void write_ignored_events_info_to_relay_log(THD *thd, Master_info *mi)
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
{
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
Relay_log_info *rli= &mi->rli;
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
pthread_mutex_t *log_lock= rli->relay_log.get_log_lock();
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("write_ignored_events_info_to_relay_log");
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
DBUG_ASSERT(thd == mi->io_thd);
pthread_mutex_lock(log_lock);
if (rli->ign_master_log_name_end[0])
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
{
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
DBUG_PRINT("info",("writing a Rotate event to track down ignored events"));
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
Rotate_log_event *ev= new Rotate_log_event(rli->ign_master_log_name_end,
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
0, rli->ign_master_log_pos_end,
Rotate_log_event::DUP_NAME);
rli->ign_master_log_name_end[0]= 0;
/* can unlock before writing as slave SQL thd will soon see our Rotate */
pthread_mutex_unlock(log_lock);
if (likely((bool)ev))
{
ev->server_id= 0; // don't be ignored by slave SQL thread
if (unlikely(rli->relay_log.append(ev)))
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
mi->report(ERROR_LEVEL, ER_SLAVE_RELAY_LOG_WRITE_FAILURE,
ER(ER_SLAVE_RELAY_LOG_WRITE_FAILURE),
"failed to write a Rotate event"
" to the relay log, SHOW SLAVE STATUS may be"
" inaccurate");
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
rli->relay_log.harvest_bytes_written(&rli->log_space_total);
many warnings (practically safe but annoying) corrected
2006-01-03 17:54:54 +01:00
if (flush_master_info(mi, 1))
sql_print_error("Failed to flush master info file");
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
delete ev;
}
else
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
mi->report(ERROR_LEVEL, ER_SLAVE_CREATE_EVENT_FAILURE,
ER(ER_SLAVE_CREATE_EVENT_FAILURE),
"Rotate_event (out of memory?),"
" SHOW SLAVE STATUS may be inaccurate");
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
}
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
else
pthread_mutex_unlock(log_lock);
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_VOID_RETURN;
Import changeset
2000-07-31 21:29:14 +02:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
int register_slave_on_master(MYSQL* mysql, Master_info *mi,
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
bool *suppress_warnings)
A slave now can optionally register with the master providing the info on how to connect to the slave for the clients that connect to the master, but would like to know where the slaves are
2001-05-31 02:50:56 +02:00
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
uchar buf[1024], *pos= buf;
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
uint report_host_len, report_user_len=0, report_password_len=0;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("register_slave_on_master");
A slave now can optionally register with the master providing the info on how to connect to the slave for the clients that connect to the master, but would like to know where the slaves are
2001-05-31 02:50:56 +02:00
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
*suppress_warnings= FALSE;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (!report_host)
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
report_host_len= strlen(report_host);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (report_user)
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
report_user_len= strlen(report_user);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
if (report_password)
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
report_password_len= strlen(report_password);
/* 30 is a good safety margin */
if (report_host_len + report_user_len + report_password_len + 30 >
sizeof(buf))
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
DBUG_RETURN(0); // safety
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
int4store(pos, server_id); pos+= 4;
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
pos= net_store_data(pos, (uchar*) report_host, report_host_len);
pos= net_store_data(pos, (uchar*) report_user, report_user_len);
pos= net_store_data(pos, (uchar*) report_password, report_password_len);
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
int2store(pos, (uint16) report_port); pos+= 2;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
int4store(pos, rpl_recovery_rank); pos+= 4;
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
/* The master will fill in master_id */
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
int4store(pos, 0); pos+= 4;
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
if (simple_command(mysql, COM_REGISTER_SLAVE, buf, (size_t) (pos- buf), 0))
A slave now can optionally register with the master providing the info on how to connect to the slave for the clients that connect to the master, but would like to know where the slaves are
2001-05-31 02:50:56 +02:00
{
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
if (mysql_errno(mysql) == ER_NET_READ_INTERRUPTED)
{
*suppress_warnings= TRUE; // Suppress reconnect warning
}
Fix for bug #29976: Excessive Slave I/O errors in replication tests Problem: The "Slave I/O thread couldn't register on master" error sporadically occurred in replication tests because the slave I/O thread got killed by STOP SLAVE before or while registering on master. Fixed by checking the state of the I/O thread, and issueing the error only if it was not explicitely killed by a user.
2007-11-22 17:22:54 +01:00
else if (!check_io_slave_killed(mi->io_thd, mi, NULL))
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
{
char buf[256];
my_snprintf(buf, sizeof(buf), "%s (Errno: %d)", mysql_error(mysql),
mysql_errno(mysql));
mi->report(ERROR_LEVEL, ER_SLAVE_MASTER_COM_FAILURE,
ER(ER_SLAVE_MASTER_COM_FAILURE), "COM_REGISTER_SLAVE", buf);
}
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
A slave now can optionally register with the master providing the info on how to connect to the slave for the clients that connect to the master, but would like to know where the slaves are
2001-05-31 02:50:56 +02:00
}
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
A slave now can optionally register with the master providing the info on how to connect to the slave for the clients that connect to the master, but would like to know where the slaves are
2001-05-31 02:50:56 +02:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
BUG#41924: high-level replication functions are not commented Adding comments to some of the high-level functions in replication.
2009-01-09 13:49:24 +01:00
/**
Execute a SHOW SLAVE STATUS statement.
@param thd Pointer to THD object for the client thread executing the
statement.
@param mi Pointer to Master_info object for the IO thread.
@retval FALSE success
@retval TRUE failure
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
bool show_master_info(THD* thd, Master_info* mi)
Import changeset
2000-07-31 21:29:14 +02:00
{
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
// TODO: fix this for multi-master
Import changeset
2000-07-31 21:29:14 +02:00
List<Item> field_list;
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
Protocol *protocol= thd->protocol;
DBUG_ENTER("show_master_info");
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
field_list.push_back(new Item_empty_string("Slave_IO_State",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
14));
Import changeset
2000-07-31 21:29:14 +02:00
field_list.push_back(new Item_empty_string("Master_Host",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
sizeof(mi->host)));
Import changeset
2000-07-31 21:29:14 +02:00
field_list.push_back(new Item_empty_string("Master_User",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
sizeof(mi->user)));
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
field_list.push_back(new Item_return_int("Master_Port", 7,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
MYSQL_TYPE_LONG));
Write slave status field names using consistent style. (Initial caps for each word.) For example, instead of writing Until_condition, Until_Log_File, and Until_log_pos, write Until_Condition, Until_Log_File, and Until_Log_pos.
2003-11-20 20:07:25 +01:00
field_list.push_back(new Item_return_int("Connect_Retry", 10,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
MYSQL_TYPE_LONG));
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
field_list.push_back(new Item_empty_string("Master_Log_File",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
FN_REFLEN));
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
field_list.push_back(new Item_return_int("Read_Master_Log_Pos", 10,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
MYSQL_TYPE_LONGLONG));
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
field_list.push_back(new Item_empty_string("Relay_Log_File",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
FN_REFLEN));
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
field_list.push_back(new Item_return_int("Relay_Log_Pos", 10,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
MYSQL_TYPE_LONGLONG));
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
field_list.push_back(new Item_empty_string("Relay_Master_Log_File",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
FN_REFLEN));
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
field_list.push_back(new Item_empty_string("Slave_IO_Running", 3));
field_list.push_back(new Item_empty_string("Slave_SQL_Running", 3));
Write slave status field names using consistent style. (Initial caps for each word.) For example, instead of writing Until_condition, Until_Log_File, and Until_log_pos, write Until_Condition, Until_Log_File, and Until_Log_pos.
2003-11-20 20:07:25 +01:00
field_list.push_back(new Item_empty_string("Replicate_Do_DB", 20));
field_list.push_back(new Item_empty_string("Replicate_Ignore_DB", 20));
field_list.push_back(new Item_empty_string("Replicate_Do_Table", 20));
field_list.push_back(new Item_empty_string("Replicate_Ignore_Table", 23));
field_list.push_back(new Item_empty_string("Replicate_Wild_Do_Table", 24));
field_list.push_back(new Item_empty_string("Replicate_Wild_Ignore_Table",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
28));
Write slave status field names using consistent style. (Initial caps for each word.) For example, instead of writing Until_condition, Until_Log_File, and Until_log_pos, write Until_Condition, Until_Log_File, and Until_Log_pos.
2003-11-20 20:07:25 +01:00
field_list.push_back(new Item_return_int("Last_Errno", 4, MYSQL_TYPE_LONG));
field_list.push_back(new Item_empty_string("Last_Error", 20));
field_list.push_back(new Item_return_int("Skip_Counter", 10,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
MYSQL_TYPE_LONG));
Write slave status field names using consistent style. (Initial caps for each word.) For example, instead of writing Until_condition, Until_Log_File, and Until_log_pos, write Until_Condition, Until_Log_File, and Until_Log_pos.
2003-11-20 20:07:25 +01:00
field_list.push_back(new Item_return_int("Exec_Master_Log_Pos", 10,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
MYSQL_TYPE_LONGLONG));
Write slave status field names using consistent style. (Initial caps for each word.) For example, instead of writing Until_condition, Until_Log_File, and Until_log_pos, write Until_Condition, Until_Log_File, and Until_Log_pos.
2003-11-20 20:07:25 +01:00
field_list.push_back(new Item_return_int("Relay_Log_Space", 10,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
MYSQL_TYPE_LONGLONG));
Write slave status field names using consistent style. (Initial caps for each word.) For example, instead of writing Until_condition, Until_Log_File, and Until_log_pos, write Until_Condition, Until_Log_File, and Until_Log_pos.
2003-11-20 20:07:25 +01:00
field_list.push_back(new Item_empty_string("Until_Condition", 6));
Implemented support for START SLAVE UNTIL (WL#868)
2003-09-13 22:13:41 +02:00
field_list.push_back(new Item_empty_string("Until_Log_File", FN_REFLEN));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
field_list.push_back(new Item_return_int("Until_Log_Pos", 10,
Implemented support for START SLAVE UNTIL (WL#868)
2003-09-13 22:13:41 +02:00
MYSQL_TYPE_LONGLONG));
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
field_list.push_back(new Item_empty_string("Master_SSL_Allowed", 7));
field_list.push_back(new Item_empty_string("Master_SSL_CA_File",
sizeof(mi->ssl_ca)));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
field_list.push_back(new Item_empty_string("Master_SSL_CA_Path",
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
sizeof(mi->ssl_capath)));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
field_list.push_back(new Item_empty_string("Master_SSL_Cert",
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
sizeof(mi->ssl_cert)));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
field_list.push_back(new Item_empty_string("Master_SSL_Cipher",
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
sizeof(mi->ssl_cipher)));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
field_list.push_back(new Item_empty_string("Master_SSL_Key",
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
sizeof(mi->ssl_key)));
Write slave status field names using consistent style. (Initial caps for each word.) For example, instead of writing Until_condition, Until_Log_File, and Until_log_pos, write Until_Condition, Until_Log_File, and Until_Log_pos.
2003-11-20 20:07:25 +01:00
field_list.push_back(new Item_return_int("Seconds_Behind_Master", 10,
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
MYSQL_TYPE_LONGLONG));
Bug#19991 CHANGE MASTER need option ssl-verify-server-cert - Add MASTER_SSL_VERIFY_SERVER_CERT option to CHANGE MASTER TO - Add Master_Ssl_Serify_Server_Cert to SHOW SLAVE STATUS - Save and restore ssl_verify_server_cert to master info file setting it to disabled as default.
2007-03-29 15:09:57 +02:00
field_list.push_back(new Item_empty_string("Master_SSL_Verify_Server_Cert",
3));
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
field_list.push_back(new Item_return_int("Last_IO_Errno", 4, MYSQL_TYPE_LONG));
field_list.push_back(new Item_empty_string("Last_IO_Error", 20));
field_list.push_back(new Item_return_int("Last_SQL_Errno", 4, MYSQL_TYPE_LONG));
field_list.push_back(new Item_empty_string("Last_SQL_Error", 20));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Port of cursors to be pushed into 5.0 tree: - client side part is simple and may be considered stable - server side part now just joggles with THD state to save execution state and has no additional locking wisdom. Lot's of it are to be rewritten.
2004-08-03 12:32:21 +02:00
if (protocol->send_fields(&field_list,
Protocol::SEND_NUM_ROWS | Protocol::SEND_EOF))
errors without code removed net_printf/send_error calls replaced by my_error family functions -1/1 (sent/unsent) error reporting removed (WL#2133)
2004-10-20 03:04:37 +02:00
DBUG_RETURN(TRUE);
Import changeset
2000-07-31 21:29:14 +02:00
Fixed timing problem with rpl000001 replication test. Fixed configure problems with HPUX and openbsd SHOW SLAVE STATUS returns empty set if slave is not initialized SHOW MASTER STATUS returns empty set if binary logging is not enabled. Fixed shutdown problem on Solaris.
2002-08-18 15:04:26 +02:00
if (mi->host[0])
{
Changed change_master() to use ER_MASTER_INFO (better display).
2003-06-17 23:19:38 +02:00
DBUG_PRINT("info",("host is set: '%s'", mi->host));
Fixed timing problem with rpl000001 replication test. Fixed configure problems with HPUX and openbsd SHOW SLAVE STATUS returns empty set if slave is not initialized SHOW MASTER STATUS returns empty set if binary logging is not enabled. Fixed shutdown problem on Solaris.
2002-08-18 15:04:26 +02:00
String *packet= &thd->packet;
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
protocol->prepare_for_resend();
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
A change of behaviour of Seconds_Behind_Master from SHOW SLAVE STATUS. It's going into 4.1 because old behaviour was somewhat nonsensical (kind of bug). Changes are that if repl threads are down or disconnected the column will be NULL, and if master is idle the column will not grow indefinitely anymore.
2004-12-16 18:12:22 +01:00
/*
Bug #34305 show slave status handling segfaults when slave io is about to leave The artifact was caused by a flaw in concurrent accessing the slave's io thd by the io itself and a handling show slave status thread. Namely, show_master_info did not acquire mi->run_lock mutex that is specified for mi->io_thd member. Fixed with deploying the mutex locking and unlocking. The mutex is kept short time and without interleaving with mi->data_lock mutex. Todo: to report and fix an issue with sys_var_slave_skip_counter::{methods} seem to acquire incorrectly active_mi->rli.run_lock instead of the specified active_mi->rli.data_lock A test case is difficult to compose, so rpl_packet should continue serving as the indicator.
2008-02-05 16:36:26 +01:00
slave_running can be accessed without run_lock but not other
non-volotile members like mi->io_thd, which is guarded by the mutex.
A change of behaviour of Seconds_Behind_Master from SHOW SLAVE STATUS. It's going into 4.1 because old behaviour was somewhat nonsensical (kind of bug). Changes are that if repl threads are down or disconnected the column will be NULL, and if master is idle the column will not grow indefinitely anymore.
2004-12-16 18:12:22 +01:00
*/
Bug #34305 show slave status handling segfaults when slave io is about to leave The artifact was caused by a flaw in concurrent accessing the slave's io thd by the io itself and a handling show slave status thread. Namely, show_master_info did not acquire mi->run_lock mutex that is specified for mi->io_thd member. Fixed with deploying the mutex locking and unlocking. The mutex is kept short time and without interleaving with mi->data_lock mutex. Todo: to report and fix an issue with sys_var_slave_skip_counter::{methods} seem to acquire incorrectly active_mi->rli.run_lock instead of the specified active_mi->rli.data_lock A test case is difficult to compose, so rpl_packet should continue serving as the indicator.
2008-02-05 16:36:26 +01:00
pthread_mutex_lock(&mi->run_lock);
protocol->store(mi->io_thd ? mi->io_thd->proc_info : "", &my_charset_bin);
pthread_mutex_unlock(&mi->run_lock);
Fixed timing problem with rpl000001 replication test. Fixed configure problems with HPUX and openbsd SHOW SLAVE STATUS returns empty set if slave is not initialized SHOW MASTER STATUS returns empty set if binary logging is not enabled. Fixed shutdown problem on Solaris.
2002-08-18 15:04:26 +02:00
pthread_mutex_lock(&mi->data_lock);
pthread_mutex_lock(&mi->rli.data_lock);
BUG#45511 rpl.rpl_binlog_corruption fails with warning messages in Valgrind This is a backport of BUG#43076.
2009-06-16 17:04:30 +02:00
pthread_mutex_lock(&mi->err_lock);
pthread_mutex_lock(&mi->rli.err_lock);
Replication were broken because of automatic charset conversion. As a fix, my_charset_bin is now used in master-slave packets instead of system_charset_info.
2003-03-18 08:34:19 +01:00
protocol->store(mi->host, &my_charset_bin);
protocol->store(mi->user, &my_charset_bin);
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
protocol->store((uint32) mi->port);
protocol->store((uint32) mi->connect_retry);
Replication were broken because of automatic charset conversion. As a fix, my_charset_bin is now used in master-slave packets instead of system_charset_info.
2003-03-18 08:34:19 +01:00
protocol->store(mi->master_log_name, &my_charset_bin);
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
protocol->store((ulonglong) mi->master_log_pos);
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
protocol->store(mi->rli.group_relay_log_name +
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
dirname_length(mi->rli.group_relay_log_name),
&my_charset_bin);
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
protocol->store((ulonglong) mi->rli.group_relay_log_pos);
protocol->store(mi->rli.group_master_log_name, &my_charset_bin);
Review of new pushed code - Fixed some error condtion when handling dates with 'T' - Added extra test for bug #11867 (Wrong result with "... WHERE ROW( a, b ) IN ( SELECT DISTINCT a, b WHERE ...)" to show it's not yet fixed - Safety fixes and cleanups
2005-08-08 23:13:49 +02:00
protocol->store(mi->slave_running == MYSQL_SLAVE_RUN_CONNECT ?
"Yes" : "No", &my_charset_bin);
Replication were broken because of automatic charset conversion. As a fix, my_charset_bin is now used in master-slave packets instead of system_charset_info.
2003-03-18 08:34:19 +01:00
protocol->store(mi->rli.slave_running ? "Yes":"No", &my_charset_bin);
WL#2472: Refactoring of replication and binlog filtering. Needed to be able to have multiple masters connected to same slave server with different filters.
2005-03-08 21:12:35 +01:00
protocol->store(rpl_filter->get_do_db());
protocol->store(rpl_filter->get_ignore_db());
On Feb 4th, 2003, Monty said he would push my patch to print replicate-do-table and other replicate-*-table options in SHOW SLAVE STATUS. Seems like it had not been done, so I push it now: there's 4 new columns to SHOW SLAVE STATUS.
2003-07-23 15:46:46 +02:00
char buf[256];
String tmp(buf, sizeof(buf), &my_charset_bin);
WL#2472: Refactoring of replication and binlog filtering. Needed to be able to have multiple masters connected to same slave server with different filters.
2005-03-08 21:12:35 +01:00
rpl_filter->get_do_table(&tmp);
On Feb 4th, 2003, Monty said he would push my patch to print replicate-do-table and other replicate-*-table options in SHOW SLAVE STATUS. Seems like it had not been done, so I push it now: there's 4 new columns to SHOW SLAVE STATUS.
2003-07-23 15:46:46 +02:00
protocol->store(&tmp);
WL#2472: Refactoring of replication and binlog filtering. Needed to be able to have multiple masters connected to same slave server with different filters.
2005-03-08 21:12:35 +01:00
rpl_filter->get_ignore_table(&tmp);
On Feb 4th, 2003, Monty said he would push my patch to print replicate-do-table and other replicate-*-table options in SHOW SLAVE STATUS. Seems like it had not been done, so I push it now: there's 4 new columns to SHOW SLAVE STATUS.
2003-07-23 15:46:46 +02:00
protocol->store(&tmp);
WL#2472: Refactoring of replication and binlog filtering. Needed to be able to have multiple masters connected to same slave server with different filters.
2005-03-08 21:12:35 +01:00
rpl_filter->get_wild_do_table(&tmp);
On Feb 4th, 2003, Monty said he would push my patch to print replicate-do-table and other replicate-*-table options in SHOW SLAVE STATUS. Seems like it had not been done, so I push it now: there's 4 new columns to SHOW SLAVE STATUS.
2003-07-23 15:46:46 +02:00
protocol->store(&tmp);
WL#2472: Refactoring of replication and binlog filtering. Needed to be able to have multiple masters connected to same slave server with different filters.
2005-03-08 21:12:35 +01:00
rpl_filter->get_wild_ignore_table(&tmp);
On Feb 4th, 2003, Monty said he would push my patch to print replicate-do-table and other replicate-*-table options in SHOW SLAVE STATUS. Seems like it had not been done, so I push it now: there's 4 new columns to SHOW SLAVE STATUS.
2003-07-23 15:46:46 +02:00
protocol->store(&tmp);
BUG#24954 (Last_errno and Last_error not set after master_retry_count has been reached): Post-merge patch to handle all the changes to the tree since the tree was cloned.
2007-06-11 22:15:39 +02:00
protocol->store(mi->rli.last_error().number);
protocol->store(mi->rli.last_error().message, &my_charset_bin);
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
protocol->store((uint32) mi->rli.slave_skip_counter);
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
protocol->store((ulonglong) mi->rli.group_master_log_pos);
rename of net_pkg.cc to protocol.cc Class for sending data from server to client (Protocol) This handles both the old ( <= 4.0 ) protocol and then new binary protocol that is used for prepared statements.
2002-12-11 08:17:51 +01:00
protocol->store((ulonglong) mi->rli.log_space_total);
Implemented support for START SLAVE UNTIL (WL#868)
2003-09-13 22:13:41 +02:00
protocol->store(
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
mi->rli.until_condition==Relay_log_info::UNTIL_NONE ? "None":
( mi->rli.until_condition==Relay_log_info::UNTIL_MASTER_POS? "Master":
Implemented support for START SLAVE UNTIL (WL#868)
2003-09-13 22:13:41 +02:00
"Relay"), &my_charset_bin);
protocol->store(mi->rli.until_log_name, &my_charset_bin);
protocol->store((ulonglong) mi->rli.until_log_pos);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
#ifdef HAVE_OPENSSL
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
protocol->store(mi->ssl? "Yes":"No", &my_charset_bin);
#else
protocol->store(mi->ssl? "Ignored":"No", &my_charset_bin);
#endif
protocol->store(mi->ssl_ca, &my_charset_bin);
protocol->store(mi->ssl_capath, &my_charset_bin);
protocol->store(mi->ssl_cert, &my_charset_bin);
protocol->store(mi->ssl_cipher, &my_charset_bin);
protocol->store(mi->ssl_key, &my_charset_bin);
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
A change of behaviour of Seconds_Behind_Master from SHOW SLAVE STATUS. It's going into 4.1 because old behaviour was somewhat nonsensical (kind of bug). Changes are that if repl threads are down or disconnected the column will be NULL, and if master is idle the column will not grow indefinitely anymore.
2004-12-16 18:12:22 +01:00
/*
Seconds_Behind_Master: if SQL thread is running and I/O thread is
connected, we can compute it otherwise show NULL (i.e. unknown).
*/
if ((mi->slave_running == MYSQL_SLAVE_RUN_CONNECT) &&
mi->rli.slave_running)
After hours of unsuccessful research on BUG#2826 "Seconds behind master weirdness" (sometimes this column of SHOW SLAVE STATUS shows a very big value, in fact a small negative number casted to ulonglong). This problem was reported by only one user, but which uses synchronized time between his servers. As suggested by the user, to hide this I display max(0, the value) so that it will be less confusing. For a user, seeing 0 is probably better than seeing -1 (both tell you that the slave is very close to the master).
2004-03-02 22:38:14 +01:00
{
Fixing portability problem by using my_time_t instead of time_t.
2007-04-13 14:56:24 +02:00
long time_diff= ((long)(time(0) - mi->rli.last_master_timestamp)
Merge mysql.com:/home/my/mysql-5.0 into mysql.com:/home/my/mysql-5.1 Merge of 'remove compiler warnings when using -Wshadow'
2007-01-27 02:46:45 +01:00
- mi->clock_diff_with_master);
After hours of unsuccessful research on BUG#2826 "Seconds behind master weirdness" (sometimes this column of SHOW SLAVE STATUS shows a very big value, in fact a small negative number casted to ulonglong). This problem was reported by only one user, but which uses synchronized time between his servers. As suggested by the user, to hide this I display max(0, the value) so that it will be less confusing. For a user, seeing 0 is probably better than seeing -1 (both tell you that the slave is very close to the master).
2004-03-02 22:38:14 +01:00
/*
Merge mysql.com:/home/my/mysql-5.0 into mysql.com:/home/my/mysql-5.1 Merge of 'remove compiler warnings when using -Wshadow'
2007-01-27 02:46:45 +01:00
Apparently on some systems time_diff can be <0. Here are possible
reasons related to MySQL:
More comments on what could explain -1 in Seconds_Behind_Master in SHOW SLAVE STATUS.
2004-03-02 23:03:52 +01:00
- the master is itself a slave of another master whose time is ahead.
- somebody used an explicit SET TIMESTAMP on the master.
Possible reason related to granularity-to-second of time functions
(nothing to do with MySQL), which can explain a value of -1:
assume the master's and slave's time are perfectly synchronized, and
that at slave's connection time, when the master's timestamp is read,
it is at the very end of second 1, and (a very short time later) when
the slave's timestamp is read it is at the very beginning of second
2. Then the recorded value for master is 1 and the recorded value for
slave is 2. At SHOW SLAVE STATUS time, assume that the difference
between timestamp of slave and rli->last_master_timestamp is 0
(i.e. they are in the same second), then we get 0-(2-1)=-1 as a result.
A change of behaviour of Seconds_Behind_Master from SHOW SLAVE STATUS. It's going into 4.1 because old behaviour was somewhat nonsensical (kind of bug). Changes are that if repl threads are down or disconnected the column will be NULL, and if master is idle the column will not grow indefinitely anymore.
2004-12-16 18:12:22 +01:00
This confuses users, so we don't go below 0: hence the max().
last_master_timestamp == 0 (an "impossible" timestamp 1970) is a
special marker to say "consider we have caught up".
After hours of unsuccessful research on BUG#2826 "Seconds behind master weirdness" (sometimes this column of SHOW SLAVE STATUS shows a very big value, in fact a small negative number casted to ulonglong). This problem was reported by only one user, but which uses synchronized time between his servers. As suggested by the user, to hide this I display max(0, the value) so that it will be less confusing. For a user, seeing 0 is probably better than seeing -1 (both tell you that the slave is very close to the master).
2004-03-02 22:38:14 +01:00
*/
Merge mysql.com:/home/my/mysql-5.0 into mysql.com:/home/my/mysql-5.1 Merge of 'remove compiler warnings when using -Wshadow'
2007-01-27 02:46:45 +01:00
protocol->store((longlong)(mi->rli.last_master_timestamp ?
max(0, time_diff) : 0));
After hours of unsuccessful research on BUG#2826 "Seconds behind master weirdness" (sometimes this column of SHOW SLAVE STATUS shows a very big value, in fact a small negative number casted to ulonglong). This problem was reported by only one user, but which uses synchronized time between his servers. As suggested by the user, to hide this I display max(0, the value) so that it will be less confusing. For a user, seeing 0 is probably better than seeing -1 (both tell you that the slave is very close to the master).
2004-03-02 22:38:14 +01:00
}
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
else
Bug#19991 CHANGE MASTER need option ssl-verify-server-cert - Add MASTER_SSL_VERIFY_SERVER_CERT option to CHANGE MASTER TO - Add Master_Ssl_Serify_Server_Cert to SHOW SLAVE STATUS - Save and restore ssl_verify_server_cert to master info file setting it to disabled as default.
2007-03-29 15:09:57 +02:00
{
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
protocol->store_null();
Bug#19991 CHANGE MASTER need option ssl-verify-server-cert - Add MASTER_SSL_VERIFY_SERVER_CERT option to CHANGE MASTER TO - Add Master_Ssl_Serify_Server_Cert to SHOW SLAVE STATUS - Save and restore ssl_verify_server_cert to master info file setting it to disabled as default.
2007-03-29 15:09:57 +02:00
}
protocol->store(mi->ssl_verify_server_cert? "Yes":"No", &my_charset_bin);
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
// Last_IO_Errno
BUG#24954 (Last_errno and Last_error not set after master_retry_count has been reached): Post-merge patch to handle all the changes to the tree since the tree was cloned.
2007-06-11 22:15:39 +02:00
protocol->store(mi->last_error().number);
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
// Last_IO_Error
BUG#24954 (Last_errno and Last_error not set after master_retry_count has been reached): Post-merge patch to handle all the changes to the tree since the tree was cloned.
2007-06-11 22:15:39 +02:00
protocol->store(mi->last_error().message, &my_charset_bin);
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
// Last_SQL_Errno
BUG#24954 (Last_errno and Last_error not set after master_retry_count has been reached): Post-merge patch to handle all the changes to the tree since the tree was cloned.
2007-06-11 22:15:39 +02:00
protocol->store(mi->rli.last_error().number);
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
// Last_SQL_Error
BUG#24954 (Last_errno and Last_error not set after master_retry_count has been reached): Post-merge patch to handle all the changes to the tree since the tree was cloned.
2007-06-11 22:15:39 +02:00
protocol->store(mi->rli.last_error().message, &my_charset_bin);
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
BUG#45511 rpl.rpl_binlog_corruption fails with warning messages in Valgrind This is a backport of BUG#43076.
2009-06-16 17:04:30 +02:00
pthread_mutex_unlock(&mi->rli.err_lock);
pthread_mutex_unlock(&mi->err_lock);
Fixed timing problem with rpl000001 replication test. Fixed configure problems with HPUX and openbsd SHOW SLAVE STATUS returns empty set if slave is not initialized SHOW MASTER STATUS returns empty set if binary logging is not enabled. Fixed shutdown problem on Solaris.
2002-08-18 15:04:26 +02:00
pthread_mutex_unlock(&mi->rli.data_lock);
pthread_mutex_unlock(&mi->data_lock);
many warnings (practically safe but annoying) corrected
2006-01-03 17:54:54 +01:00
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
if (my_net_write(&thd->net, (uchar*) thd->packet.ptr(), packet->length()))
errors without code removed net_printf/send_error calls replaced by my_error family functions -1/1 (sent/unsent) error reporting removed (WL#2133)
2004-10-20 03:04:37 +02:00
DBUG_RETURN(TRUE);
Fixed timing problem with rpl000001 replication test. Fixed configure problems with HPUX and openbsd SHOW SLAVE STATUS returns empty set if slave is not initialized SHOW MASTER STATUS returns empty set if binary logging is not enabled. Fixed shutdown problem on Solaris.
2002-08-18 15:04:26 +02:00
}
Rename send_eof() to my_eof() for consistency with my_ok() and my_error()
2008-02-19 13:58:08 +01:00
my_eof(thd);
errors without code removed net_printf/send_error calls replaced by my_error family functions -1/1 (sent/unsent) error reporting removed (WL#2133)
2004-10-20 03:04:37 +02:00
DBUG_RETURN(FALSE);
Import changeset
2000-07-31 21:29:14 +02:00
}
rpl000016.test sync rpl000001.result BitKeeper file /home/sasha/src/bk/mysql/mysql-test/r/rpl000001.result ignore Added BitKeeper/tmp/bkr3sAHD to the ignore list slave.h MASTER_POS_WAIT lex.h MASTER_POS_WAIT slave.cc MASTER_POS_WAIT, do automagic restart on debugging abort, skip rotate events in slave.cc debug abort count sql_repl.cc announce the log name at the start of the log with a fake rotate event item_create.h MASTER_POS_WAIT item_func.cc MASTER_POS_WAIT item_func.h MASTER_POS_WAIT sql_class.h enter_cond(), exit_cond() helper inliners item_create.cc added MASTER_POS_WAIT mysql-test-run.sh speed improvement fixes rpl000007.test sync rpl000003.test sleep -> sync rpl000004.test sleep -> sync, fixed clean up bug rpl000014.test sync rpl000009.test sync rpl000013.test sync rpl000001.test sleep -> sync rpl000008.test sync rpl000006.test sync on cleanup rpl000011.test sync rpl000012.test sync rpl000005.test sleep -> sync rpl000010.test sync rpl000015.test sync rpl000002.test sleep -> sync rpl000014.result we now know the master log name as soon as we connect mysql.cc added optional agrument to --wait mysqltest.c added save_master_pos and sync_with_master commands
2001-01-17 13:47:33 +01:00
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
void set_slave_thread_options(THD* thd)
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("set_slave_thread_options");
Fix for bug #19736 VIEW: column names not quoted properly when view is replicated When we write 'query=...' string to a frm file for views on a slave, indentifiers are not properly quoted due to missing OPTION_QUOTE_SHOW_CREATE flag in the thd->options. Fix: properly set thd->options for the slave thread.
2006-11-03 12:27:37 +01:00
/*
It's nonsense to constrain the slave threads with max_join_size; if a
query succeeded on master, we HAVE to execute it. So set
OPTION_BIG_SELECTS. Setting max_join_size to HA_POS_ERROR is not enough
(and it's not needed if we have OPTION_BIG_SELECTS) because an INSERT
SELECT examining more than 4 billion rows would still fail (yes, because
when max_join_size is 4G, OPTION_BIG_SELECTS is automatically set, but
only for client threads.
*/
ulonglong options= thd->options | OPTION_BIG_SELECTS;
if (opt_log_slave_updates)
options|= OPTION_BIN_LOG;
else
options&= ~OPTION_BIN_LOG;
thd->options= options;
Ensure that slaves have normal COMMIT behaviour after WL#1967
2005-02-07 14:15:48 +01:00
thd->variables.completion_type= 0;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_VOID_RETURN;
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
void set_slave_thread_default_charset(THD* thd, Relay_log_info const *rli)
Last part of WL#1062: better replication of timezones: no more use of SET ONE_SHOT; storing tz info directly in event (if this info is needed), it's now allowed to have different global tz on master and slave.
2005-03-22 00:26:12 +01:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("set_slave_thread_default_charset");
Last part of WL#1062: better replication of timezones: no more use of SET ONE_SHOT; storing tz info directly in event (if this info is needed), it's now allowed to have different global tz on master and slave.
2005-03-22 00:26:12 +01:00
thd->variables.character_set_client=
global_system_variables.character_set_client;
thd->variables.collation_connection=
global_system_variables.collation_connection;
thd->variables.collation_server=
global_system_variables.collation_server;
thd->update_charset();
BUG#23171 (Illegal slave restart group position): Second patch to fix skipping code. Moving relay and binary log position changing code from do_apply_event [old exec_event()] into do_update_pos() and doing other changes necessary to support that. Fixing a bug that can cause deadlock if rotating binary log when committing a changes to a transactional table that is not inside a transaction and cause a rotate log.
2006-11-10 15:10:41 +01:00
/*
We use a const cast here since the conceptual (and externally
visible) behavior of the function is to set the default charset of
the thread. That the cache has to be invalidated is a secondary
effect.
*/
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
const_cast<Relay_log_info*>(rli)->cached_charset_invalidate();
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_VOID_RETURN;
Last part of WL#1062: better replication of timezones: no more use of SET ONE_SHOT; storing tz info directly in event (if this info is needed), it's now allowed to have different global tz on master and slave.
2005-03-22 00:26:12 +01:00
}
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
/*
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
init_slave_thread()
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
*/
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
static int init_slave_thread(THD* thd, SLAVE_THD_TYPE thd_type)
Import changeset
2000-07-31 21:29:14 +02:00
{
DBUG_ENTER("init_slave_thread");
Bug #33931 assertion at write_ignored_events_info_to_relay_log if init_slave_thread() fails and bug#33932 assertion at handle_slave_sql if init_slave_thread() fails the asserts were caused by bug33931: having thd deleted at time of executing err: code plus a missed initialization; bug33932: initialization of slave_is_running member was missed; fixed with relocating mi members initialization and removing delete thd It is safe to do as deletion happens later explicitly in the caller of init_slave_thread(). Todo: at merging the test is better to be moved into suite/bugs for 5.x (when x>0).
2008-02-13 13:09:41 +01:00
#if !defined(DBUG_OFF)
int simulate_error= 0;
#endif
- Fix for BUG#1858 "SQL-Thread stops working when using optimize table": we change THD::system_thread from a 'bool' to a bitmap to be able to distinguish between delayed-insert threads and slave threads. - Fix for BUG#1701 "Update from multiple tables" (one line in sql_parse.cc, plus a new test rpl_multi_update.test). That's just adding an initialization.
2003-12-04 22:42:18 +01:00
thd->system_thread = (thd_type == SLAVE_THD_SQL) ?
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
SYSTEM_THREAD_SLAVE_SQL : SYSTEM_THREAD_SLAVE_IO;
WL#2787 (part 2, ver 3 (merged)) changed securety context switching
2005-09-15 21:29:07 +02:00
thd->security_ctx->skip_grants();
Import changeset
2000-07-31 21:29:14 +02:00
my_net_init(&thd->net, 0);
bug#19402 SQL close to the size of the max_allowed_packet fails on the slave comments are fixed as was suggested in reviews.
2006-11-12 19:01:58 +01:00
/*
Adding MAX_LOG_EVENT_HEADER_LEN to the max_allowed_packet on all
slave threads, since a replication event can become this much larger
than the corresponding packet (query) sent from client to master.
*/
BUG#19402 SQL close to the size of the max_allowed_packet fails on the slave A communication packet can also be a binlog event sent from the master to the slave. To be sent by master dump and accepted by slave io thread both have to have the value of max_allowed_packet bigger than one that client connection had. In the patch there is the MAX possible replicatio header size estimation for events in binlog that embedded user query. Only these events of query_log_event type, i.e just plain queries, require attention.
2006-09-11 23:19:05 +02:00
thd->variables.max_allowed_packet= global_system_variables.max_allowed_packet
bug#19402 SQL close to the size of the max_allowed_packet fails on the slave comments are fixed as was suggested in reviews.
2006-11-12 19:01:58 +01:00
+ MAX_LOG_EVENT_HEADER; /* note, incr over the global not session var */
repl-tests/test-repl-ts/repl-timestamp.master.reject this file needs to be deleted sql/log_event.cc fixed warnings sql/log_event.h fixed () #define bug sql/mysqlbinlog.cc fixed length argument - was supposed to be one less sql/mysqld.cc replicate-do/ignore-table now works, wild does not yet sql/mysql_priv.h updating argument to add_table_to_list() -- needed for replicate-do/ignore table sql/slave.cc changes fore replicate-do/ignore-table close the socket before going to sleep sleeping after error bad event was being freed too early sql/slave.h changes for replicate-do/ignore-table sql/sql_class.cc slave_thread variable to THD sql/sql_class.h added slave_thread to THD, fixed bug in end_time() sql/sql_parse.cc updating argument to add_tables_to_list() sql/sql_table.cc fixed bug in mysql_rm_table() sql/sql_yacc.yy fixed up add_table_to_list() calls to accept updating argument sql/table.h added updating to TABLE_LIST
2000-11-14 07:43:02 +01:00
thd->slave_thread = 1;
Slow query log to file now displays queries with microsecond precission --long-query-time is now given in seconds with microseconds as decimals --min_examined_row_limit added for slow query log long_query_time user variable is now double with 6 decimals Added functions to get time in microseconds Added faster time() functions for system that has gethrtime() (Solaris) We now do less time() calls. Added field->in_read_set() and field->in_write_set() for easier field manipulation by handlers set_var.cc and my_getopt() can now handle DOUBLE variables. All time() calls changed to my_time() my_time() now does retry's if time() call fails. Added debug function for stopping in mysql_admin_table() when tables are locked Some trivial function and struct variable renames to avoid merge errors. Fixed compiler warnings Initialization of some time variables on windows moved to my_init()
2007-07-30 10:33:50 +02:00
thd->enable_slow_log= opt_log_slow_slave_statements;
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
set_slave_thread_options(thd);
Import changeset
2000-07-31 21:29:14 +02:00
thd->client_capabilities = CLIENT_LOCAL_FILES;
pthread_mutex_lock(&LOCK_thread_count);
Fixed compiler warnings Fixed compile-pentium64 scripts Fixed wrong estimate of update_with_key_prefix in sql-bench Merge bk-internal.mysql.com:/home/bk/mysql-5.1 into mysql.com:/home/my/mysql-5.1 Fixed unsafe define of uint4korr() Fixed that --extern works with mysql-test-run.pl Small trivial cleanups This also fixes a bug in counting number of rows that are updated when we have many simultanous queries Move all connection handling and command exectuion main loop from sql_parse.cc to sql_connection.cc Split handle_one_connection() into reusable sub functions. Split create_new_thread() into reusable sub functions. Added thread_scheduler; Preliminary interface code for future thread_handling code. Use 'my_thread_id' for internal thread id's Make thr_alarm_kill() to depend on thread_id instead of thread Make thr_abort_locks_for_thread() depend on thread_id instead of thread In store_globals(), set my_thread_var->id to be thd->thread_id. Use my_thread_var->id as basis for my_thread_name() The above changes makes the connection we have between THD and threads more soft. Added a lot of DBUG_PRINT() and DBUG_ASSERT() functions Fixed compiler warnings Fixed core dumps when running with --debug Removed setting of signal masks (was never used) Made event code call pthread_exit() (portability fix) Fixed that event code doesn't call DBUG_xxx functions before my_thread_init() is called. Made handling of thread_id and thd->variables.pseudo_thread_id uniform. Removed one common 'not freed memory' warning from mysqltest Fixed a couple of usage of not initialized warnings (unlikely cases) Suppress compiler warnings from bdb and (for the moment) warnings from ndb
2007-02-23 12:13:55 +01:00
thd->thread_id= thd->variables.pseudo_thread_id= thread_id++;
Import changeset
2000-07-31 21:29:14 +02:00
pthread_mutex_unlock(&LOCK_thread_count);
Bug #33931 assertion at write_ignored_events_info_to_relay_log if init_slave_thread() fails and bug#33932 assertion at handle_slave_sql if init_slave_thread() fails the asserts were caused by bug33931: having thd deleted at time of executing err: code plus a missed initialization; bug33932: initialization of slave_is_running member was missed; fixed with relocating mi members initialization and removing delete thd It is safe to do as deletion happens later explicitly in the caller of init_slave_thread(). Todo: at merging the test is better to be moved into suite/bugs for 5.x (when x>0).
2008-02-13 13:09:41 +01:00
DBUG_EXECUTE_IF("simulate_io_slave_error_on_init",
simulate_error|= (1 << SLAVE_THD_IO););
DBUG_EXECUTE_IF("simulate_sql_slave_error_on_init",
simulate_error|= (1 << SLAVE_THD_SQL););
#if !defined(DBUG_OFF)
if (init_thr_lock() || thd->store_globals() || simulate_error & (1<< thd_type))
#else
replication updates. This changeset seems to be working fine on test systems. If no problems are discovered in the next week, this will make the replication code ready for 4.0.2 release.
2002-03-30 20:36:05 +01:00
if (init_thr_lock() || thd->store_globals())
Bug #33931 assertion at write_ignored_events_info_to_relay_log if init_slave_thread() fails and bug#33932 assertion at handle_slave_sql if init_slave_thread() fails the asserts were caused by bug33931: having thd deleted at time of executing err: code plus a missed initialization; bug33932: initialization of slave_is_running member was missed; fixed with relocating mi members initialization and removing delete thd It is safe to do as deletion happens later explicitly in the caller of init_slave_thread(). Todo: at merging the test is better to be moved into suite/bugs for 5.x (when x>0).
2008-02-13 13:09:41 +01:00
#endif
Import changeset
2000-07-31 21:29:14 +02:00
{
Fix that round(0.1,1) == round(0.1,1)
2003-03-10 11:00:19 +01:00
thd->cleanup();
Import changeset
2000-07-31 21:29:14 +02:00
DBUG_RETURN(-1);
}
Bug#31210 - INSERT DELAYED crashes server when used on partitioned table Trying INSERT DELAYED on a partitioned table, that has not been used right before, crashes the server. When a table is used for select or update, it is kept open for some time. This period I mean with "right before". Information about partitioning of a table is stored in form of a string in the .frm file. Parsing of this string requires a correctly set up lexical analyzer (lex). The partitioning code uses a new temporary instance of a lex. But it does still refer to the previously active lex. The delayd insert thread does not initialize its lex though... Added initialization for thd->lex before open table in the delayed thread and at all other places where it is necessary to call lex_start() if all tables would be partitioned and need to parse the .frm file.
2007-11-05 16:25:40 +01:00
lex_start(thd);
Import changeset
2000-07-31 21:29:14 +02:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (thd_type == SLAVE_THD_SQL)
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Waiting for the next event in relay log");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
else
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Waiting for master update");
Import changeset
2000-07-31 21:29:14 +02:00
thd->version=refresh_version;
thd->set_time();
DBUG_RETURN(0);
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
fixed replication assert failure ( found by Jeremy Zawodny)
2002-03-08 23:02:11 +01:00
static int safe_sleep(THD* thd, int sec, CHECK_KILLED_FUNC thread_killed,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
void* thread_killed_arg)
Import changeset
2000-07-31 21:29:14 +02:00
{
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
int nap_time;
Import changeset
2000-07-31 21:29:14 +02:00
thr_alarm_t alarmed;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("safe_sleep");
Import changeset
2000-07-31 21:29:14 +02:00
thr_alarm_init(&alarmed);
Slow query log to file now displays queries with microsecond precission --long-query-time is now given in seconds with microseconds as decimals --min_examined_row_limit added for slow query log long_query_time user variable is now double with 6 decimals Added functions to get time in microseconds Added faster time() functions for system that has gethrtime() (Solaris) We now do less time() calls. Added field->in_read_set() and field->in_write_set() for easier field manipulation by handlers set_var.cc and my_getopt() can now handle DOUBLE variables. All time() calls changed to my_time() my_time() now does retry's if time() call fails. Added debug function for stopping in mysql_admin_table() when tables are locked Some trivial function and struct variable renames to avoid merge errors. Fixed compiler warnings Initialization of some time variables on windows moved to my_init()
2007-07-30 10:33:50 +02:00
time_t start_time= my_time(0);
Import changeset
2000-07-31 21:29:14 +02:00
time_t end_time= start_time+sec;
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
while ((nap_time= (int) (end_time - start_time)) > 0)
Import changeset
2000-07-31 21:29:14 +02:00
{
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
ALARM alarm_buff;
keep bk happy
2001-01-22 03:46:32 +01:00
/*
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
The only reason we are asking for alarm is so that
keep bk happy
2001-01-22 03:46:32 +01:00
we will be woken up in case of murder, so if we do not get killed,
set the alarm so it goes off after we wake up naturally
*/
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
thr_alarm(&alarmed, 2 * nap_time, &alarm_buff);
Import changeset
2000-07-31 21:29:14 +02:00
sleep(nap_time);
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
thr_end_alarm(&alarmed);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
fixed replication assert failure ( found by Jeremy Zawodny)
2002-03-08 23:02:11 +01:00
if ((*thread_killed)(thd,thread_killed_arg))
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
Slow query log to file now displays queries with microsecond precission --long-query-time is now given in seconds with microseconds as decimals --min_examined_row_limit added for slow query log long_query_time user variable is now double with 6 decimals Added functions to get time in microseconds Added faster time() functions for system that has gethrtime() (Solaris) We now do less time() calls. Added field->in_read_set() and field->in_write_set() for easier field manipulation by handlers set_var.cc and my_getopt() can now handle DOUBLE variables. All time() calls changed to my_time() my_time() now does retry's if time() call fails. Added debug function for stopping in mysql_admin_table() when tables are locked Some trivial function and struct variable renames to avoid merge errors. Fixed compiler warnings Initialization of some time variables on windows moved to my_init()
2007-07-30 10:33:50 +02:00
start_time= my_time(0);
Import changeset
2000-07-31 21:29:14 +02:00
}
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
Import changeset
2000-07-31 21:29:14 +02:00
}
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int request_dump(MYSQL* mysql, Master_info* mi,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
bool *suppress_warnings)
Import changeset
2000-07-31 21:29:14 +02:00
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
uchar buf[FN_REFLEN + 10];
Import changeset
2000-07-31 21:29:14 +02:00
int len;
int binlog_flags = 0; // for now
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
char* logname = mi->master_log_name;
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
DBUG_ENTER("request_dump");
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
*suppress_warnings= FALSE;
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
// TODO if big log files: Change next to int8store()
More debugging Print position in normal log for Binlog dump
2004-11-10 17:07:39 +01:00
int4store(buf, (ulong) mi->master_log_pos);
Import changeset
2000-07-31 21:29:14 +02:00
int2store(buf + 4, binlog_flags);
bug fixes in server-id, moved replication functionality to sql_repl.* will push this one, since the replication code now works
2000-09-30 01:20:26 +02:00
int4store(buf + 6, server_id);
FLUSH TABLE table_list Fixes for RENAME TABLE Portability fixes
2000-08-21 23:18:32 +02:00
len = (uint) strlen(logname);
bug fixes in server-id, moved replication functionality to sql_repl.* will push this one, since the replication code now works
2000-09-30 01:20:26 +02:00
memcpy(buf + 10, logname,len);
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
if (simple_command(mysql, COM_BINLOG_DUMP, buf, len + 10, 1))
keep bk happy
2001-01-22 03:46:32 +01:00
{
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
Something went wrong, so we will just reconnect and retry later
in the future, we should do a better error analysis, but for
now we just fill up the error log :-)
*/
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
if (mysql_errno(mysql) == ER_NET_READ_INTERRUPTED)
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
*suppress_warnings= TRUE; // Suppress reconnect warning
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
else
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
sql_print_error("Error on COM_BINLOG_DUMP: %d %s, will retry in %d secs",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mysql_errno(mysql), mysql_error(mysql),
master_connect_retry);
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
DBUG_RETURN(1);
keep bk happy
2001-01-22 03:46:32 +01:00
}
Import changeset
2000-07-31 21:29:14 +02:00
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
DBUG_RETURN(0);
Import changeset
2000-07-31 21:29:14 +02:00
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
updates for LOAD DATA FROM MASTER + some cleanup of replication code
2001-05-29 03:18:23 +02:00
static int request_table_dump(MYSQL* mysql, const char* db, const char* table)
Import changeset
2000-07-31 21:29:14 +02:00
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
uchar buf[1024], *p = buf;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("request_table_dump");
FLUSH TABLE table_list Fixes for RENAME TABLE Portability fixes
2000-08-21 23:18:32 +02:00
uint table_len = (uint) strlen(table);
uint db_len = (uint) strlen(db);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
if (table_len + db_len > sizeof(buf) - 2)
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
{
sql_print_error("request_table_dump: Buffer overrun");
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
}
Import changeset
2000-07-31 21:29:14 +02:00
*p++ = db_len;
memcpy(p, db, db_len);
p += db_len;
*p++ = table_len;
memcpy(p, table, table_len);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
if (simple_command(mysql, COM_TABLE_DUMP, buf, p - buf + table_len, 1))
keep bk happy
2001-01-22 03:46:32 +01:00
{
sql_print_error("request_table_dump: Error sending the table dump \
Import changeset
2000-07-31 21:29:14 +02:00
command");
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
keep bk happy
2001-01-22 03:46:32 +01:00
}
Import changeset
2000-07-31 21:29:14 +02:00
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
Import changeset
2000-07-31 21:29:14 +02:00
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
/*
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Read one event from the master
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
SYNOPSIS
read_event()
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mysql MySQL connection
mi Master connection information
suppress_warnings TRUE when a normal net read timeout has caused us to
try a reconnect. We do not want to print anything to
the error log in this case because this a anormal
event in an idle server.
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
RETURN VALUES
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
'packet_error' Error
number Length of packet
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static ulong read_event(MYSQL* mysql, Master_info *mi, bool* suppress_warnings)
Import changeset
2000-07-31 21:29:14 +02:00
{
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
ulong len;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("read_event");
keep bk happy
2001-01-22 03:46:32 +01:00
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
*suppress_warnings= FALSE;
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
my_real_read() will time us out
We check if we were told to die, and if not, try reading again
*/
mysql-test/mysql-test-run cleanup - remove unneeded daeamon restarts sql/log_event.cc fixed bug - do not confuse confused slaves even further if they try to read less than LOG_EVENT_HEADER_LEN before the end of file sql/mysqlbinlog.cc give error if constructor fails sql/mysqld.cc --disconnect-slave-event-count for replication debugging sql/slave.cc --disconnect-slave-event-count cleanup to remove unneeded sleeps when things go wrong - sleep only when we have to sql/slave.h --disconnect-slave-event-count test cases for ALTER TABLE ORDER BY and replication offset off by 22 confusion mystery bug (both pass)
2000-11-22 08:23:31 +01:00
#ifndef DBUG_OFF
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
if (disconnect_slave_event_count && !(mi->events_till_disconnect--))
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
DBUG_RETURN(packet_error);
mysql-test/mysql-test-run cleanup - remove unneeded daeamon restarts sql/log_event.cc fixed bug - do not confuse confused slaves even further if they try to read less than LOG_EVENT_HEADER_LEN before the end of file sql/mysqlbinlog.cc give error if constructor fails sql/mysqld.cc --disconnect-slave-event-count for replication debugging sql/slave.cc --disconnect-slave-event-count cleanup to remove unneeded sleeps when things go wrong - sleep only when we have to sql/slave.h --disconnect-slave-event-count test cases for ALTER TABLE ORDER BY and replication offset off by 22 confusion mystery bug (both pass)
2000-11-22 08:23:31 +01:00
#endif
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
A fix and a test case for Bug#15752 "Lost connection to MySQL server when calling a SP from C API" The bug was caused by lack of checks for misuse in mysql_real_query. A stored procedure always returns at least one result, which is the status of execution of the procedure itself. This result, or so-called OK packet, is similar to a result returned by INSERT/UPDATE/CREATE operations: it contains the overall status of execution, the number of affected rows and the number of warnings. The client test program attached to the bug did not read this result and ivnoked the next query. In turn, libmysql had no check for such scenario and mysql_real_query was simply trying to send that query without reading the pending response, thus messing up the communication protocol. The fix is to return an error from mysql_real_query when it's called prior to retrieval of all pending results.
2006-07-24 12:56:53 +02:00
len = cli_safe_read(mysql);
64-bit portability fixes in network I/O use $MAX_TABLES mysqltest variable to make join test portable make test now works with the manager on IA64 Linux
2001-10-13 17:28:35 +02:00
if (len == packet_error || (long) len < 1)
Import changeset
2000-07-31 21:29:14 +02:00
{
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
if (mysql_errno(mysql) == ER_NET_READ_INTERRUPTED)
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
{
/*
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
We are trying a normal reconnect after a read timeout;
we suppress prints to .err file as long as the reconnect
happens without problems
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
*/
*suppress_warnings= TRUE;
}
else
auto-ROLLBACK if binlog was not closed properly auto-commit on Xid_log_event
2005-02-09 20:04:28 +01:00
sql_print_error("Error reading packet from server: %s ( server_errno=%d)",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mysql_error(mysql), mysql_errno(mysql));
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(packet_error);
Import changeset
2000-07-31 21:29:14 +02:00
}
hanged UDF interface to use clear() instead of reset()
2003-08-27 21:30:50 +02:00
/* Check if eof packet */
if (len < 8 && mysql->net.read_pos[0] == 254)
keep bk happy
2001-01-22 03:46:32 +01:00
{
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
sql_print_information("Slave: received end packet from server, apparent "
"master shutdown: %s",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mysql_error(mysql));
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(packet_error);
keep bk happy
2001-01-22 03:46:32 +01:00
}
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Fixed a LOT of compiler warnings Added missing DBUG_RETURN statements (in mysqldump.c) Added missing enums Fixed a lot of wrong DBUG_PRINT() statements, some of which could cause crashes Removed usage of %lld and %p in printf strings as these are not portable or produces different results on different systems.
2006-11-27 00:47:38 +01:00
DBUG_PRINT("exit", ("len: %lu net->read_pos[4]: %d",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
len, mysql->net.read_pos[4]));
DBUG_RETURN(len - 1);
Import changeset
2000-07-31 21:29:14 +02:00
}
wl2325 wl2324
2006-01-12 19:51:02 +01:00
/*
Check if the current error is of temporary nature of not.
Some errors are temporary in nature, such as
ER_LOCK_DEADLOCK and ER_LOCK_WAIT_TIMEOUT. Ndb also signals
that the error is temporary by pushing a warning with the error code
ER_GET_TEMPORARY_ERRMSG, if the originating error is temporary.
*/
static int has_temporary_error(THD *thd)
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("has_temporary_error");
BUG#24860 (Incorrect SLAVE_TRANSACTION_RETRIES code can result in slave stuck): If a temporary error occured inside a group on an event that was not the first event of the group, the slave could get stuck because the retry counter is reset whenever an event was executed successfully. This patch only reset the retry counter when an entire group has been successfully executed, or failed with a non-transient error.
2007-10-20 20:16:12 +02:00
DBUG_EXECUTE_IF("all_errors_are_temporary_errors",
Bug#12713 "Error in a stored function called from a SELECT doesn't cause ROLLBACK of statement", part 1. Review fixes. Do not send OK/EOF packets to the client until we reached the end of the current statement. This is a consolidation, to keep the functionality that is shared by all SQL statements in one place in the server. Currently this functionality includes: - close_thread_tables() - log_slow_statement(). After this patch and the subsequent patch for Bug#12713, it shall also include: - ha_autocommit_or_rollback() - net_end_statement() - query_cache_end_of_result(). In future it may also include: - mysql_reset_thd_for_next_command().
2007-12-12 16:21:01 +01:00
if (thd->main_da.is_error())
{
thd->clear_error();
my_error(ER_LOCK_DEADLOCK, MYF(0));
});
/*
If there is no message in THD, we can't say if it's a temporary
error or not. This is currently the case for Incident_log_event,
which sets no message. Return FALSE.
*/
if (!thd->is_error())
DBUG_RETURN(0);
wl2325 wl2324
2006-01-12 19:51:02 +01:00
/*
Temporary error codes:
currently, InnoDB deadlock detected by InnoDB or lock
wait timeout (innodb_lock_wait_timeout exceeded
*/
Bug#12713 "Error in a stored function called from a SELECT doesn't cause ROLLBACK of statement", part 1. Review fixes. Do not send OK/EOF packets to the client until we reached the end of the current statement. This is a consolidation, to keep the functionality that is shared by all SQL statements in one place in the server. Currently this functionality includes: - close_thread_tables() - log_slow_statement(). After this patch and the subsequent patch for Bug#12713, it shall also include: - ha_autocommit_or_rollback() - net_end_statement() - query_cache_end_of_result(). In future it may also include: - mysql_reset_thd_for_next_command().
2007-12-12 16:21:01 +01:00
if (thd->main_da.sql_errno() == ER_LOCK_DEADLOCK ||
thd->main_da.sql_errno() == ER_LOCK_WAIT_TIMEOUT)
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
wl2325 wl2324
2006-01-12 19:51:02 +01:00
#ifdef HAVE_NDB_BINLOG
/*
currently temporary error set in ndbcluster
*/
List_iterator_fast<MYSQL_ERROR> it(thd->warn_list);
MYSQL_ERROR *err;
while ((err= it++))
{
dbug changes: 1. dbug state is now local to a thread 2. new macros: DBUG_EXPLAIN, DBUG_EXPLAIN_INITIAL, DBUG_SET, DBUG_SET_INITIAL, DBUG_EVALUATE, DBUG_EVALUATE_IF 3. macros are do{}while(0) wrapped 4. incremental modifications to the dbug state (e.g. "+d,info:-t") 5. dbug code cleanup, style fixes 6. _db_on_ and DEBUGGER_ON/OFF removed 7. rest of MySQL code fixed because of 3 (missing ;) and 6 8. dbug manual updated 9. server variable @@debug (global and local) to control dbug from SQL! a. -#T to print timestamps in the log
2006-02-14 22:36:11 +01:00
DBUG_PRINT("info", ("has warning %d %s", err->code, err->msg));
wl2325 wl2324
2006-01-12 19:51:02 +01:00
switch (err->code)
{
case ER_GET_TEMPORARY_ERRMSG:
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
wl2325 wl2324
2006-01-12 19:51:02 +01:00
default:
break;
}
}
#endif
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
wl2325 wl2324
2006-01-12 19:51:02 +01:00
}
WL#1062 "log charset info into all Query_log_event": we store 7 bytes (1 + 2*3) in every Query_log_event. In the future if users want binlog optimized for small size and less safe, we could add --binlog-no-charset (and binlog-no-sql-mode etc): charset info is something by design optional (even if for now we don't offer possibility to disable it): it's not a binlog format change. We try to reduce the number of get_charset() calls in the slave SQL thread to a minimum by caching the charset read from the previous event (which will often be equal to the one of the current event). We don't use SET ONE_SHOT for charset-aware repl (we still do for timezones, will be fixed later). No more errors if one changes the global value of charset vars on master or slave (as we log charset info in all Query_log_event). Not fixing Load_log_event as it will be rewritten soon by Dmitri. Testing how mysqlbinlog behaves in rpl_charset.test. mysqlbinlog needs to know where charset file is (to be able to convert a charset number found in binlog (e.g. in User_var_log_event) to a charset name); mysql-test-run needs to pass the correct value for this option to mysqlbinlog. Many result udpates (adding charset info into every event shifts log_pos in SHOW BINLOG EVENTS). Roughly the same job is to be done for timezones :)
2005-02-03 16:22:16 +01:00
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
/**
Applies the given event and advances the relay log position.
In essence, this function does:
@code
ev->apply_event(rli);
ev->update_pos(rli);
@endcode
But it also does some maintainance, such as skipping events if
needed and reporting errors.
If the @c skip flag is set, then it is tested whether the event
should be skipped, by looking at the slave_skip_counter and the
server id. The skip flag should be set when calling this from a
replication thread but not set when executing an explicit BINLOG
statement.
@retval 0 OK.
@retval 1 Error calling ev->apply_event().
@retval 2 No error calling ev->apply_event(), but error calling
ev->update_pos().
*/
Bug#46640: output from mysqlbinlog command in 5.1 breaks replication The BINLOG statement was sharing too much code with the slave SQL thread, introduced with the patch for Bug#32407. This caused statements to be logged with the wrong server_id, the id stored inside the events of the BINLOG statement rather than the id of the running server. Fix by rearranging code a bit so that only relevant parts of the code are executed by the BINLOG statement, and the server_id of the server executing the statements will not be overrided by the server_id stored in the 'format description BINLOG statement'.
2009-10-14 03:39:05 +02:00
int apply_event_and_update_pos(Log_event* ev, THD* thd, Relay_log_info* rli)
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
{
int exec_res= 0;
DBUG_ENTER("apply_event_and_update_pos");
DBUG_PRINT("exec_event",("%s(type_code: %d; server_id: %d)",
ev->get_type_str(), ev->get_type_code(),
ev->server_id));
DBUG_PRINT("info", ("thd->options: %s%s; rli->last_event_start_time: %lu",
FLAGSTR(thd->options, OPTION_NOT_AUTOCOMMIT),
FLAGSTR(thd->options, OPTION_BEGIN),
rli->last_event_start_time));
/*
Execute the event to change the database and update the binary
log coordinates, but first we set some data that is needed for
the thread.
The event will be executed unless it is supposed to be skipped.
Queries originating from this server must be skipped. Low-level
events (Format_description_log_event, Rotate_log_event,
Stop_log_event) from this server must also be skipped. But for
those we don't want to modify 'group_master_log_pos', because
these events did not exist on the master.
Format_description_log_event is not completely skipped.
Skip queries specified by the user in 'slave_skip_counter'. We
can't however skip events that has something to do with the log
files themselves.
Filtering on own server id is extremely important, to ignore
execution of events created by the creation/rotation of the relay
log (remember that now the relay log starts with its Format_desc,
has a Rotate etc).
*/
thd->server_id = ev->server_id; // use the original server id for logging
thd->set_time(); // time the query
thd->lex->current_select= 0;
if (!ev->when)
ev->when= my_time(0);
ev->thd = thd; // because up to this point, ev->thd == 0
Bug#46640: output from mysqlbinlog command in 5.1 breaks replication The BINLOG statement was sharing too much code with the slave SQL thread, introduced with the patch for Bug#32407. This caused statements to be logged with the wrong server_id, the id stored inside the events of the BINLOG statement rather than the id of the running server. Fix by rearranging code a bit so that only relevant parts of the code are executed by the BINLOG statement, and the server_id of the server executing the statements will not be overrided by the server_id stored in the 'format description BINLOG statement'.
2009-10-14 03:39:05 +02:00
int reason= ev->shall_skip(rli);
if (reason == Log_event::EVENT_SKIP_COUNT)
--rli->slave_skip_counter;
pthread_mutex_unlock(&rli->data_lock);
if (reason == Log_event::EVENT_SKIP_NOT)
exec_res= ev->apply_event(rli);
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
#ifndef DBUG_OFF
Bug#46640: output from mysqlbinlog command in 5.1 breaks replication The BINLOG statement was sharing too much code with the slave SQL thread, introduced with the patch for Bug#32407. This caused statements to be logged with the wrong server_id, the id stored inside the events of the BINLOG statement rather than the id of the running server. Fix by rearranging code a bit so that only relevant parts of the code are executed by the BINLOG statement, and the server_id of the server executing the statements will not be overrided by the server_id stored in the 'format description BINLOG statement'.
2009-10-14 03:39:05 +02:00
/*
This only prints information to the debug trace.
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
Bug#46640: output from mysqlbinlog command in 5.1 breaks replication The BINLOG statement was sharing too much code with the slave SQL thread, introduced with the patch for Bug#32407. This caused statements to be logged with the wrong server_id, the id stored inside the events of the BINLOG statement rather than the id of the running server. Fix by rearranging code a bit so that only relevant parts of the code are executed by the BINLOG statement, and the server_id of the server executing the statements will not be overrided by the server_id stored in the 'format description BINLOG statement'.
2009-10-14 03:39:05 +02:00
TODO: Print an informational message to the error log?
*/
static const char *const explain[] = {
// EVENT_SKIP_NOT,
"not skipped",
// EVENT_SKIP_IGNORE,
"skipped because event should be ignored",
// EVENT_SKIP_COUNT
"skipped because event skip counter was non-zero"
};
DBUG_PRINT("info", ("OPTION_BEGIN: %d; IN_STMT: %d",
thd->options & OPTION_BEGIN ? 1 : 0,
rli->get_flag(Relay_log_info::IN_STMT)));
DBUG_PRINT("skip_event", ("%s event was %s",
ev->get_type_str(), explain[reason]));
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
#endif
DBUG_PRINT("info", ("apply_event error = %d", exec_res));
if (exec_res == 0)
{
int error= ev->update_pos(rli);
Patch to eliminate some valgrind warnings in debug printout code.
2008-02-06 14:44:47 +01:00
#ifdef HAVE_purify
if (!rli->is_fake)
#endif
{
Disabling declaration of debug variable for non-debug builds.
2008-02-07 08:41:32 +01:00
#ifndef DBUG_OFF
Patch to eliminate some valgrind warnings in debug printout code.
2008-02-06 14:44:47 +01:00
char buf[22];
Disabling declaration of debug variable for non-debug builds.
2008-02-07 08:41:32 +01:00
#endif
Patch to eliminate some valgrind warnings in debug printout code.
2008-02-06 14:44:47 +01:00
DBUG_PRINT("info", ("update_pos error = %d", error));
DBUG_PRINT("info", ("group %s %s",
llstr(rli->group_relay_log_pos, buf),
rli->group_relay_log_name));
DBUG_PRINT("info", ("event %s %s",
llstr(rli->event_relay_log_pos, buf),
rli->event_relay_log_name));
}
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
/*
The update should not fail, so print an error message and
return an error code.
TODO: Replace this with a decent error message when merged
with BUG#24954 (which adds several new error message).
*/
if (error)
{
Patch to eliminate some valgrind warnings in debug printout code.
2008-02-06 14:44:47 +01:00
char buf[22];
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
rli->report(ERROR_LEVEL, ER_UNKNOWN_ERROR,
"It was not possible to update the positions"
" of the relay log information: the slave may"
" be in an inconsistent state."
" Stopped in %s position %s",
rli->group_relay_log_name,
llstr(rli->group_relay_log_pos, buf));
DBUG_RETURN(2);
}
}
DBUG_RETURN(exec_res ? 1 : 0);
}
/**
Top-level function for executing the next event from the relay log.
This function reads the event from the relay log, executes it, and
advances the relay log position. It also handles errors, etc.
This function may fail to apply the event for the following reasons:
- The position specfied by the UNTIL condition of the START SLAVE
command is reached.
- It was not possible to read the event from the log.
- The slave is killed.
- An error occurred when applying the event, and the event has been
tried slave_trans_retries times. If the event has been retried
fewer times, 0 is returned.
- init_master_info or init_relay_log_pos failed. (These are called
BUG#41924: high-level replication functions are not commented Adding comments to some of the high-level functions in replication.
2009-01-09 13:49:24 +01:00
if a failure occurs when applying the event.)
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
- An error occurred when updating the binlog position.
@retval 0 The event was applied.
@retval 1 The event was not applied.
*/
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
static int exec_relay_log_event(THD* thd, Relay_log_info* rli)
Import changeset
2000-07-31 21:29:14 +02:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("exec_relay_log_event");
Implemented support for START SLAVE UNTIL (WL#868)
2003-09-13 22:13:41 +02:00
/*
We acquire this mutex since we need it for all operations except
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
event execution. But we will release it in places where we will
Implemented support for START SLAVE UNTIL (WL#868)
2003-09-13 22:13:41 +02:00
wait for something for example inside of next_event().
*/
pthread_mutex_lock(&rli->data_lock);
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
Log_event * ev = next_event(rli);
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(rli->sql_thd==thd);
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
fixed replication assert failure ( found by Jeremy Zawodny)
2002-03-08 23:02:11 +01:00
if (sql_slave_killed(thd,rli))
- Fix for memory leak if the SQL slave thread is killed just after reading an event. - A few more mutex locks and unlocks of rli.log_space_lock for doing clean reads of rli.ignore_log_space_limit - Broadcast after unlock, not before (small speed optimisation).
2003-05-25 23:09:46 +02:00
{
Implemented support for START SLAVE UNTIL (WL#868)
2003-09-13 22:13:41 +02:00
pthread_mutex_unlock(&rli->data_lock);
Code/testcase cleanups
2003-07-01 12:29:55 +02:00
delete ev;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
- Fix for memory leak if the SQL slave thread is killed just after reading an event. - A few more mutex locks and unlocks of rli.log_space_lock for doing clean reads of rli.ignore_log_space_limit - Broadcast after unlock, not before (small speed optimisation).
2003-05-25 23:09:46 +02:00
}
Fixed indention, removed compiler varnings and fixed a bug in FULLTEXT indexes.
2000-09-28 23:58:16 +02:00
if (ev)
{
LOAD DATA INFILE is now replicated properly, except for cleanup on Stop event and bugs the test suite could not catch Did some big restructuring of binlog event classes - most important change is that now each event class has exec_event method and one does not need to modify slave core code to add a new event. Slave code is now much smaller and easier to read
2001-08-03 23:57:53 +02:00
int exec_res;
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
BUG#13861 - START SLAVE UNTIL may stop 1 evnt too late if log-slave-updates and circul repl Slave SQL thread may execute one extra event when there are events skipped by slave I/O thread (e.g. originated by the same server). Whereas it was requested not to do so by the UNTIL condition. This happens because we compare with the end position of previously executed event. This is fine when there are no skipped by slave I/O thread events, as end position of previous event equals to start position of to be executed event. Otherwise this position equals to start position of skipped event. This is fixed by: - reading the event to be executed before checking if the until condition is satisfied. - comparing the start position of the event to be executed. Since we do not have the start position available, we compute it by subtracting event length from end position (which is available). - if there are no events on the event queue at the slave sql starting time, that meet until condition, we stop immediately, as in this case we do not want to wait for next event.
2008-02-22 16:07:07 +01:00
/*
This tests if the position of the beginning of the current event
hits the UNTIL barrier.
*/
BUG#13861 - START SLAVE UNTIL may stop 1 evnt too late if log-slave-updates and circul repl After merge fixes.
2008-02-27 18:46:06 +01:00
if (rli->until_condition != Relay_log_info::UNTIL_NONE &&
Merge mysql.com:/home/svoj/devel/mysql/BUG13861/mysql-5.0-engines into mysql.com:/home/svoj/devel/mysql/BUG13861/mysql-5.1-engines
2008-02-27 14:43:54 +01:00
rli->is_until_satisfied((rli->is_in_group() || !ev->log_pos) ?
BUG#13861 - START SLAVE UNTIL may stop 1 evnt too late if log-slave-updates and circul repl Slave SQL thread may execute one extra event when there are events skipped by slave I/O thread (e.g. originated by the same server). Whereas it was requested not to do so by the UNTIL condition. This happens because we compare with the end position of previously executed event. This is fine when there are no skipped by slave I/O thread events, as end position of previous event equals to start position of to be executed event. Otherwise this position equals to start position of skipped event. This is fixed by: - reading the event to be executed before checking if the until condition is satisfied. - comparing the start position of the event to be executed. Since we do not have the start position available, we compute it by subtracting event length from end position (which is available). - if there are no events on the event queue at the slave sql starting time, that meet until condition, we stop immediately, as in this case we do not want to wait for next event.
2008-02-22 16:07:07 +01:00
rli->group_master_log_pos :
ev->log_pos - ev->data_written))
{
char buf[22];
sql_print_information("Slave SQL thread stopped because it reached its"
" UNTIL position %s", llstr(rli->until_pos(), buf));
/*
Setting abort_slave flag because we do not want additional message about
error in query execution to be printed.
*/
rli->abort_slave= 1;
pthread_mutex_unlock(&rli->data_lock);
delete ev;
Merge mysql.com:/home/svoj/devel/mysql/BUG13861/mysql-5.0-engines into mysql.com:/home/svoj/devel/mysql/BUG13861/mysql-5.1-engines
2008-02-27 14:43:54 +01:00
DBUG_RETURN(1);
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
}
Bug#46640: output from mysqlbinlog command in 5.1 breaks replication The BINLOG statement was sharing too much code with the slave SQL thread, introduced with the patch for Bug#32407. This caused statements to be logged with the wrong server_id, the id stored inside the events of the BINLOG statement rather than the id of the running server. Fix by rearranging code a bit so that only relevant parts of the code are executed by the BINLOG statement, and the server_id of the server executing the statements will not be overrided by the server_id stored in the 'format description BINLOG statement'.
2009-10-14 03:39:05 +02:00
exec_res= apply_event_and_update_pos(ev, thd, rli);
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
BUG#23171 (Illegal slave restart position): Third patch of the bug fix where the code for skipping events and for executing events is factored out into three functions: - shall_skip() to decide if the event shall be skipped and the reason for it; - do_apply_event(), where the event is applied to the database; and - do_update_pos(), which updates the actual relay log position and group positions.
2007-01-17 15:06:37 +01:00
/*
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
Format_description_log_event should not be deleted because it will be
used to read info about the relay log's format; it will be deleted when
the SQL thread does not need it, i.e. when this thread terminates.
BUG#23171 (Illegal slave restart position): Third patch of the bug fix where the code for skipping events and for executing events is factored out into three functions: - shall_skip() to decide if the event shall be skipped and the reason for it; - do_apply_event(), where the event is applied to the database; and - do_update_pos(), which updates the actual relay log position and group positions.
2007-01-17 15:06:37 +01:00
*/
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
if (ev->get_type_code() != FORMAT_DESCRIPTION_EVENT)
BUG#23171: Illegal group log position Tail fixes after re-applying patches to older version of clone.
2007-03-22 08:32:41 +01:00
{
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
DBUG_PRINT("info", ("Deleting the event after it has been executed"));
delete ev;
BUG#23171: Illegal group log position Tail fixes after re-applying patches to older version of clone.
2007-03-22 08:32:41 +01:00
}
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
/*
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
update_log_pos failed: this should not happen, so we don't
retry.
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
*/
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
if (exec_res == 2)
DBUG_RETURN(1);
Fix for BUG#8325 "Deadlock in replication thread stops replication": in slave SQL thread: if a transaction fails because of InnoDB deadlock or innodb_lock_wait_timeout exceeded, optionally retry the transaction a certain number of times (new variable --slave_transaction_retries).
2005-03-02 11:29:48 +01:00
if (slave_trans_retries)
{
BUG#24860 (Incorrect SLAVE_TRANSACTION_RETRIES code can result in slave stuck): If a temporary error occured inside a group on an event that was not the first event of the group, the slave could get stuck because the retry counter is reset whenever an event was executed successfully. This patch only reset the retry counter when an entire group has been successfully executed, or failed with a non-transient error.
2007-10-20 20:16:12 +02:00
int temp_err;
if (exec_res && (temp_err= has_temporary_error(thd)))
Fix for BUG#8325 "Deadlock in replication thread stops replication": in slave SQL thread: if a transaction fails because of InnoDB deadlock or innodb_lock_wait_timeout exceeded, optionally retry the transaction a certain number of times (new variable --slave_transaction_retries).
2005-03-02 11:29:48 +01:00
{
const char *errmsg;
/*
We were in a transaction which has been rolled back because of a
wl2325 wl2324
2006-01-12 19:51:02 +01:00
temporary error;
let's seek back to BEGIN log event and retry it all again.
Merge dsl-hkigw8-feaef900-46.dhcp.inet.fi:/home/elkin/MySQL/TEAM/BARE/5.0-merge into dsl-hkigw8-feaef900-46.dhcp.inet.fi:/home/elkin/MySQL/TEAM/FIXES/merge_50
2006-10-30 12:07:36 +01:00
Note, if lock wait timeout (innodb_lock_wait_timeout exceeded)
BUG#20697 slave fails to rollback replicated transaction hang over innodb_lock_wait_timeou Transaction on the slave sql thread got blocked against a slave's mysqld local ta's lock. Since the default, slave-transaction-retries=10, there was replaying of the replicated ta. That failed because of a new started from 5.0.13 policy not to rollback a timed-out transaction. Effectively the first round of a timed-out ta becomes committed by the replaying's first "BEGIN". It was decided to backport already existed method working in 5.1 implemented in bug #16228 for handling symmetrical deadlock problem. That patch introduced end_trans execution whenever a replicated ta deadlocks or timed-out. Note, that this solution can be practically suboptimal - in the light of the changed behavior due to timeout we still could replay only the last statement - only with a high rate of timeouting replicated transactions.
2006-10-11 09:16:37 +02:00
there is no rollback since 5.0.13 (ref: manual).
Fix for BUG#8325 "Deadlock in replication thread stops replication": in slave SQL thread: if a transaction fails because of InnoDB deadlock or innodb_lock_wait_timeout exceeded, optionally retry the transaction a certain number of times (new variable --slave_transaction_retries).
2005-03-02 11:29:48 +01:00
We have to not only seek but also
a) init_master_info(), to seek back to hot relay log's start for later
(for when we will come back to this hot log after re-processing the
possibly existing old logs where BEGIN is: check_binlog_magic() will
then need the cache to be at position 0 (see comments at beginning of
init_master_info()).
b) init_relay_log_pos(), because the BEGIN may be an older relay log.
*/
"After Monty's review" changes to the fix for BUG#8325 "Deadlock in replication thread stops replication": s/sleep/safe_sleep (thread safe); sleep 0/1/2/3/4/5/5/5 (get slave less late); no message on error log (deadlock is too common sometimes), a global counter instead (SHOW STATUS LIKE 'slave_retried_transactions'). Plus a fix for libmysql/Makefile.shared
2005-03-23 19:19:36 +01:00
if (rli->trans_retries < slave_trans_retries)
Fix for BUG#8325 "Deadlock in replication thread stops replication": in slave SQL thread: if a transaction fails because of InnoDB deadlock or innodb_lock_wait_timeout exceeded, optionally retry the transaction a certain number of times (new variable --slave_transaction_retries).
2005-03-02 11:29:48 +01:00
{
if (init_master_info(rli->mi, 0, 0, 0, SLAVE_SQL))
sql_print_error("Failed to initialize the master info structure");
else if (init_relay_log_pos(rli,
rli->group_relay_log_name,
rli->group_relay_log_pos,
First part of fix for BUG#7998 "Replication should be more clever about when to replicate RELEASE_LOCK()" (next part is updating test results)
2005-03-02 16:37:54 +01:00
1, &errmsg, 1))
Fix for BUG#8325 "Deadlock in replication thread stops replication": in slave SQL thread: if a transaction fails because of InnoDB deadlock or innodb_lock_wait_timeout exceeded, optionally retry the transaction a certain number of times (new variable --slave_transaction_retries).
2005-03-02 11:29:48 +01:00
sql_print_error("Error initializing relay log position: %s",
errmsg);
else
{
exec_res= 0;
Bug#16228 (RBR: Slave SQL thread retries infinitely): Transaction aborted on slave should be retries. The OPTION_BEGIN bit was cleared prematurely. Removed dependence of code on value of OPTION_BEGIN bit when executing retries.
2006-03-07 13:29:53 +01:00
end_trans(thd, ROLLBACK);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
/* chance for concurrent connection to get more locks */
"After Monty's review" changes to the fix for BUG#8325 "Deadlock in replication thread stops replication": s/sleep/safe_sleep (thread safe); sleep 0/1/2/3/4/5/5/5 (get slave less late); no message on error log (deadlock is too common sometimes), a global counter instead (SHOW STATUS LIKE 'slave_retried_transactions'). Plus a fix for libmysql/Makefile.shared
2005-03-23 19:19:36 +01:00
safe_sleep(thd, min(rli->trans_retries, MAX_SLAVE_RETRY_PAUSE),
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
(CHECK_KILLED_FUNC)sql_slave_killed, (void*)rli);
"After Monty's review" changes to the fix for BUG#8325 "Deadlock in replication thread stops replication": s/sleep/safe_sleep (thread safe); sleep 0/1/2/3/4/5/5/5 (get slave less late); no message on error log (deadlock is too common sometimes), a global counter instead (SHOW STATUS LIKE 'slave_retried_transactions'). Plus a fix for libmysql/Makefile.shared
2005-03-23 19:19:36 +01:00
pthread_mutex_lock(&rli->data_lock); // because of SHOW STATUS
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
rli->trans_retries++;
"After Monty's review" changes to the fix for BUG#8325 "Deadlock in replication thread stops replication": s/sleep/safe_sleep (thread safe); sleep 0/1/2/3/4/5/5/5 (get slave less late); no message on error log (deadlock is too common sometimes), a global counter instead (SHOW STATUS LIKE 'slave_retried_transactions'). Plus a fix for libmysql/Makefile.shared
2005-03-23 19:19:36 +01:00
rli->retried_trans++;
pthread_mutex_unlock(&rli->data_lock);
DBUG_PRINT("info", ("Slave retries transaction "
"rli->trans_retries: %lu", rli->trans_retries));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
}
Fix for BUG#8325 "Deadlock in replication thread stops replication": in slave SQL thread: if a transaction fails because of InnoDB deadlock or innodb_lock_wait_timeout exceeded, optionally retry the transaction a certain number of times (new variable --slave_transaction_retries).
2005-03-02 11:29:48 +01:00
}
else
sql_print_error("Slave SQL thread retried transaction %lu time(s) "
"in vain, giving up. Consider raising the value of "
"the slave_transaction_retries variable.",
slave_trans_retries);
}
Merge from 5.0-bugteam for 43414
2009-06-09 18:44:26 +02:00
else if ((exec_res && !temp_err) ||
BUG#24860 (Incorrect SLAVE_TRANSACTION_RETRIES code can result in slave stuck): If a temporary error occured inside a group on an event that was not the first event of the group, the slave could get stuck because the retry counter is reset whenever an event was executed successfully. This patch only reset the retry counter when an entire group has been successfully executed, or failed with a non-transient error.
2007-10-20 20:16:12 +02:00
(opt_using_transactions &&
rli->group_relay_log_pos == rli->event_relay_log_pos))
Bug#16228 (RBR: Slave SQL thread retries infinitely): Transaction aborted on slave should be retries. The OPTION_BEGIN bit was cleared prematurely. Removed dependence of code on value of OPTION_BEGIN bit when executing retries.
2006-03-07 13:29:53 +01:00
{
/*
BUG#24860 (Incorrect SLAVE_TRANSACTION_RETRIES code can result in slave stuck): If a temporary error occured inside a group on an event that was not the first event of the group, the slave could get stuck because the retry counter is reset whenever an event was executed successfully. This patch only reset the retry counter when an entire group has been successfully executed, or failed with a non-transient error.
2007-10-20 20:16:12 +02:00
Only reset the retry counter if the entire group succeeded
or failed with a non-transient error. On a successful
event, the execution will proceed as usual; in the case of a
Bug#16228 (RBR: Slave SQL thread retries infinitely): Transaction aborted on slave should be retries. The OPTION_BEGIN bit was cleared prematurely. Removed dependence of code on value of OPTION_BEGIN bit when executing retries.
2006-03-07 13:29:53 +01:00
non-transient error, the slave will stop with an error.
*/
rli->trans_retries= 0; // restart from fresh
Manual merge 5.0-rpl -> 5.1-rpl. involved bug#12691, bug#27571
2007-10-30 12:49:42 +01:00
DBUG_PRINT("info", ("Resetting retry counter, rli->trans_retries: %lu",
BUG#24860 (Incorrect SLAVE_TRANSACTION_RETRIES code can result in slave stuck): If a temporary error occured inside a group on an event that was not the first event of the group, the slave could get stuck because the retry counter is reset whenever an event was executed successfully. This patch only reset the retry counter when an entire group has been successfully executed, or failed with a non-transient error.
2007-10-20 20:16:12 +02:00
rli->trans_retries));
Bug#16228 (RBR: Slave SQL thread retries infinitely): Transaction aborted on slave should be retries. The OPTION_BEGIN bit was cleared prematurely. Removed dependence of code on value of OPTION_BEGIN bit when executing retries.
2006-03-07 13:29:53 +01:00
}
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
}
DBUG_RETURN(exec_res);
Fixed indention, removed compiler varnings and fixed a bug in FULLTEXT indexes.
2000-09-28 23:58:16 +02:00
}
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
pthread_mutex_unlock(&rli->data_lock);
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
rli->report(ERROR_LEVEL, ER_SLAVE_RELAY_LOG_READ_FAILURE,
ER(ER_SLAVE_RELAY_LOG_READ_FAILURE), "\
Clearer error message (in the customer's case, the relay log was corrupted, not the master's binlog) (SW 1571).
2003-06-02 17:30:47 +02:00
Could not parse relay log event entry. The possible reasons are: the master's \
binary log is corrupted (you can check this by running 'mysqlbinlog' on the \
binary log), the slave's relay log is corrupted (you can check this by running \
'mysqlbinlog' on the relay log), a network problem, or a bug in the master's \
or slave's MySQL code. If you want to check the master's binary log or slave's \
relay log, you will be able to know their names by issuing 'SHOW SLAVE STATUS' \
on this slave.\
Cleanups...
2001-02-21 03:39:48 +01:00
");
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
Import changeset
2000-07-31 21:29:14 +02:00
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static bool check_io_slave_killed(THD *thd, Master_info *mi, const char *info)
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
{
if (io_slave_killed(thd, mi))
{
Fix for bug #29976: Excessive Slave I/O errors in replication tests Problem: The "Slave I/O thread couldn't register on master" error sporadically occurred in replication tests because the slave I/O thread got killed by STOP SLAVE before or while registering on master. Fixed by checking the state of the I/O thread, and issueing the error only if it was not explicitely killed by a user.
2007-11-22 17:22:54 +01:00
if (info && global_system_variables.log_warnings)
Bug #43414 Parenthesis (and other) warnings compiling MySQL with gcc 4.3.2 Cleaning up warnings not present in 5.0.
2009-09-23 15:21:29 +02:00
sql_print_information("%s", info);
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
return TRUE;
}
return FALSE;
}
/**
@brief Try to reconnect slave IO thread.
@details Terminates current connection to master, sleeps for
@c mi->connect_retry msecs and initiates new connection with
@c safe_reconnect(). Variable pointed by @c retry_count is increased -
if it exceeds @c master_retry_count then connection is not re-established
and function signals error.
Unless @c suppres_warnings is TRUE, a warning is put in the server error log
when reconnecting. The warning message and messages used to report errors
are taken from @c messages array. In case @c master_retry_count is exceeded,
no messages are added to the log.
@param[in] thd Thread context.
@param[in] mysql MySQL connection.
@param[in] mi Master connection information.
@param[in,out] retry_count Number of attempts to reconnect.
@param[in] suppress_warnings TRUE when a normal net read timeout
has caused to reconnecting.
@param[in] messages Messages to print/log, see
reconnect_messages[] array.
@retval 0 OK.
@retval 1 There was an error.
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int try_to_reconnect(THD *thd, MYSQL *mysql, Master_info *mi,
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
uint *retry_count, bool suppress_warnings,
const char *messages[SLAVE_RECON_MSG_MAX])
{
mi->slave_running= MYSQL_SLAVE_RUN_NOT_CONNECT;
thd->proc_info= messages[SLAVE_RECON_MSG_WAIT];
#ifdef SIGNAL_WITH_VIO_CLOSE
thd->clear_active_vio();
#endif
end_server(mysql);
if ((*retry_count)++)
{
if (*retry_count > master_retry_count)
return 1; // Don't retry forever
safe_sleep(thd, mi->connect_retry, (CHECK_KILLED_FUNC) io_slave_killed,
(void *) mi);
}
if (check_io_slave_killed(thd, mi, messages[SLAVE_RECON_MSG_KILLED_WAITING]))
return 1;
thd->proc_info = messages[SLAVE_RECON_MSG_AFTER];
if (!suppress_warnings)
{
char buf[256], llbuff[22];
my_snprintf(buf, sizeof(buf), messages[SLAVE_RECON_MSG_FAILED],
IO_RPL_LOG_NAME, llstr(mi->master_log_pos, llbuff));
/*
Raise a warining during registering on master/requesting dump.
Log a message reading event.
*/
if (messages[SLAVE_RECON_MSG_COMMAND][0])
{
mi->report(WARNING_LEVEL, ER_SLAVE_MASTER_COM_FAILURE,
ER(ER_SLAVE_MASTER_COM_FAILURE),
messages[SLAVE_RECON_MSG_COMMAND], buf);
}
else
{
Bug #43414 Parenthesis (and other) warnings compiling MySQL with gcc 4.3.2 Cleaning up warnings not present in 5.0.
2009-09-23 15:21:29 +02:00
sql_print_information("%s", buf);
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
}
}
if (safe_reconnect(thd, mysql, mi, 1) || io_slave_killed(thd, mi))
{
if (global_system_variables.log_warnings)
Bug #43414 Parenthesis (and other) warnings compiling MySQL with gcc 4.3.2 Cleaning up warnings not present in 5.0.
2009-09-23 15:21:29 +02:00
sql_print_information("%s", messages[SLAVE_RECON_MSG_KILLED_AFTER]);
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
return 1;
}
return 0;
}
BUG#41924: high-level replication functions are not commented Adding comments to some of the high-level functions in replication.
2009-01-09 13:49:24 +01:00
/**
Slave IO thread entry point.
@param arg Pointer to Master_info struct that holds information for
the IO thread.
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
BUG#41924: high-level replication functions are not commented Adding comments to some of the high-level functions in replication.
2009-01-09 13:49:24 +01:00
@return Always 0.
*/
pthread_handler_decl() changed to be ctags-friendly (and contain extern "C" when necessary)
2005-10-08 16:39:55 +02:00
pthread_handler_t handle_slave_io(void *arg)
Import changeset
2000-07-31 21:29:14 +02:00
{
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
THD *thd; // needs to be first for thread_stack
MYSQL *mysql;
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
Master_info *mi = (Master_info*)arg;
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
Relay_log_info *rli= &mi->rli;
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
char llbuff[22];
uint retry_count;
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
bool suppress_warnings;
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
int ret;
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
#ifndef DBUG_OFF
uint retry_count_reg= 0, retry_count_dump= 0, retry_count_event= 0;
#endif
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
// needs to call my_thread_init(), otherwise we get a coredump in DBUG_ stuff
my_thread_init();
Fix that round(0.1,1) == round(0.1,1)
2003-03-10 11:00:19 +01:00
DBUG_ENTER("handle_slave_io");
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(mi->inited);
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
mysql= NULL ;
retry_count= 0;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_mutex_lock(&mi->run_lock);
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
/* Inform waiting threads that slave has started */
mi->slave_run_id++;
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
#ifndef DBUG_OFF
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
mi->events_till_disconnect = disconnect_slave_event_count;
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
#endif
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
thd= new THD; // note that contructor of THD uses DBUG_ !
replication updates. This changeset seems to be working fine on test systems. If no problems are discovered in the next week, this will make the replication code ready for 4.0.2 release.
2002-03-30 20:36:05 +01:00
THD_CHECK_SENTRY(thd);
Bug #33931 assertion at write_ignored_events_info_to_relay_log if init_slave_thread() fails and bug#33932 assertion at handle_slave_sql if init_slave_thread() fails the asserts were caused by bug33931: having thd deleted at time of executing err: code plus a missed initialization; bug33932: initialization of slave_is_running member was missed; fixed with relocating mi members initialization and removing delete thd It is safe to do as deletion happens later explicitly in the caller of init_slave_thread(). Todo: at merging the test is better to be moved into suite/bugs for 5.x (when x>0).
2008-02-13 13:09:41 +01:00
mi->io_thd = thd;
Import changeset
2000-07-31 21:29:14 +02:00
pthread_detach_this_thread();
Always initialize THD::thread_stack: it's used in check_stack_overrun().
2005-11-23 19:18:10 +01:00
thd->thread_stack= (char*) &thd; // remember where our stack is
BUG#44270: RESET SLAVE does not reset Last_IO_Error or Last_IO_Errno The server was not cleaning the last IO error and error number when resetting slave. This patch addresses this issue by backporting into 5.1 part of the patch in BUG 34654. A fix for this issue had already been pushed into 6.0 as part of the aforementioned bug, however the patch also included some refactoring. The fix for 5.1 does not take into account the refactoring part.
2009-06-03 16:14:18 +02:00
mi->clear_error();
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (init_slave_thread(thd, SLAVE_THD_IO))
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
{
pthread_cond_broadcast(&mi->start_cond);
pthread_mutex_unlock(&mi->run_lock);
sql_print_error("Failed during slave I/O thread initialization");
goto err;
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
pthread_mutex_lock(&LOCK_thread_count);
Import changeset
2000-07-31 21:29:14 +02:00
threads.append(thd);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
pthread_mutex_unlock(&LOCK_thread_count);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
mi->slave_running = 1;
mi->abort_slave = 0;
pthread_mutex_unlock(&mi->run_lock);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
pthread_cond_broadcast(&mi->start_cond);
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
DBUG_PRINT("master_info",("log_file_name: '%s' position: %s",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mi->master_log_name,
llstr(mi->master_log_pos,llbuff)));
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
if (!(mi->mysql = mysql = mysql_init(NULL)))
keep bk happy
2001-01-22 03:46:32 +01:00
{
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
mi->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
ER(ER_SLAVE_FATAL_ERROR), "error in mysql_init()");
keep bk happy
2001-01-22 03:46:32 +01:00
goto err;
}
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Connecting to master");
added loops and other cool stuff to mysqltest fixed slave to preserve temp tables on slave stop; slave start added abort-slave-event-count option to mysqld to test the above added a test case for temp tables with a contantly aborting slave removed warning in sql_parse.cc fixed thimble.cc to compile
2000-12-02 18:11:50 +01:00
// we can get killed during safe_connect
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (!safe_connect(thd, mysql, mi))
BUG#19402 SQL close to the size of the max_allowed_packet fails on the slave A communication packet can also be a binlog event sent from the master to the slave. To be sent by master dump and accepted by slave io thread both have to have the value of max_allowed_packet bigger than one that client connection had. In the patch there is the MAX possible replicatio header size estimation for events in binlog that embedded user query. Only these events of query_log_event type, i.e just plain queries, require attention.
2006-09-11 23:19:05 +02:00
{
Fixed a LOT of compiler warnings Added missing DBUG_RETURN statements (in mysqldump.c) Added missing enums Fixed a lot of wrong DBUG_PRINT() statements, some of which could cause crashes Removed usage of %lld and %p in printf strings as these are not portable or produces different results on different systems.
2006-11-27 00:47:38 +01:00
sql_print_information("Slave I/O thread: connected to master '%s@%s:%d',"
"replication started in log '%s' at position %s",
mi->user, mi->host, mi->port,
BUG#19402 SQL close to the size of the max_allowed_packet fails on the slave A communication packet can also be a binlog event sent from the master to the slave. To be sent by master dump and accepted by slave io thread both have to have the value of max_allowed_packet bigger than one that client connection had. In the patch there is the MAX possible replicatio header size estimation for events in binlog that embedded user query. Only these events of query_log_event type, i.e just plain queries, require attention.
2006-09-11 23:19:05 +02:00
IO_RPL_LOG_NAME,
llstr(mi->master_log_pos,llbuff));
bug#19402 SQL close to the size of the max_allowed_packet fails on the slave comments are fixed as was suggested in reviews.
2006-11-12 19:01:58 +01:00
/*
Adding MAX_LOG_EVENT_HEADER_LEN to the max_packet_size on the I/O
thread, since a replication event can become this much larger than
the corresponding packet (query) sent from client to master.
*/
BUG#19402 SQL close to the size of the max_allowed_packet fails on the slave A communication packet can also be a binlog event sent from the master to the slave. To be sent by master dump and accepted by slave io thread both have to have the value of max_allowed_packet bigger than one that client connection had. In the patch there is the MAX possible replicatio header size estimation for events in binlog that embedded user query. Only these events of query_log_event type, i.e just plain queries, require attention.
2006-09-11 23:19:05 +02:00
mysql->net.max_packet_size= thd->net.max_packet_size+= MAX_LOG_EVENT_HEADER;
}
added loops and other cool stuff to mysqltest fixed slave to preserve temp tables on slave stop; slave start added abort-slave-event-count option to mysqld to test the above added a test case for temp tables with a contantly aborting slave removed warning in sql_parse.cc fixed thimble.cc to compile
2000-12-02 18:11:50 +01:00
else
Fixed test cases for innobase Fixed bug introduced with last ORDER BY optimization Changed log position to longlong to avoid warnings.
2001-03-07 22:50:44 +01:00
{
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
sql_print_information("Slave I/O thread killed while connecting to master");
Fixed test cases for innobase Fixed bug introduced with last ORDER BY optimization Changed log position to longlong to avoid warnings.
2001-03-07 22:50:44 +01:00
goto err;
}
A slave now can optionally register with the master providing the info on how to connect to the slave for the clients that connect to the master, but would like to know where the slaves are
2001-05-31 02:50:56 +02:00
fixed message in post-commit trigger added debugging options to master to behave sporadically, and a test case to make sure the slave can deal with it
2001-06-29 02:22:29 +02:00
connected:
merged with 3.23 replication updates
2001-06-29 03:48:49 +02:00
A change of behaviour of Seconds_Behind_Master from SHOW SLAVE STATUS. It's going into 4.1 because old behaviour was somewhat nonsensical (kind of bug). Changes are that if repl threads are down or disconnected the column will be NULL, and if master is idle the column will not grow indefinitely anymore.
2004-12-16 18:12:22 +01:00
// TODO: the assignment below should be under mutex (5.0)
mi->slave_running= MYSQL_SLAVE_RUN_CONNECT;
LOAD DATA INFILE is now replicated properly, except for cleanup on Stop event and bugs the test suite could not catch Did some big restructuring of binlog event classes - most important change is that now each event class has exec_event method and one does not need to modify slave core code to add a new event. Slave code is now much smaller and easier to read
2001-08-03 23:57:53 +02:00
thd->slave_net = &mysql->net;
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Checking master version");
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
ret= get_master_version_and_clock(mysql, mi);
if (ret == 1)
/* Fatal error */
work to enable reading 3.23 logs - not yet finished moved fail-safe replication routines from sql_repl.cc to repl_failsafe.cc write start event only in the first log
2001-11-11 06:24:12 +01:00
goto err;
Bug #45214 get_master_version_and_clock does not report error when queries fail The "get_master_version_and_clock(...)" function in sql/slave.cc ignores error and passes directly when queries fail, or queries succeed but the result retrieved is empty. The "get_master_version_and_clock(...)" function should try to reconnect master if queries fail because of transient network problems, and fail otherwise. The I/O thread should print a warning if the some system variables do not exist on master (very old master)
2009-07-16 08:56:43 +02:00
if (ret == 2)
{
if (check_io_slave_killed(mi->io_thd, mi, "Slave I/O thread killed"
"while calling get_master_version_and_clock(...)"))
goto err;
suppress_warnings= FALSE;
/* Try to reconnect because the error was caused by a transient network problem */
if (try_to_reconnect(thd, mysql, mi, &retry_count, suppress_warnings,
reconnect_messages[SLAVE_RECON_ACT_REG]))
goto err;
goto connected;
}
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
if (mi->rli.relay_log.description_event_for_queue->binlog_version > 1)
work to enable reading 3.23 logs - not yet finished moved fail-safe replication routines from sql_repl.cc to repl_failsafe.cc write start event only in the first log
2001-11-11 06:24:12 +01:00
{
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
/*
Register ourselves with the master.
*/
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Registering slave on master");
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
if (register_slave_on_master(mysql, mi, &suppress_warnings))
BUG#21123 (Slave fails to reconnect on update_slave_list): Fixing comments and adding error message if slave fails to register on master.
2007-05-31 10:17:16 +02:00
{
Fix for bug #29976: Excessive Slave I/O errors in replication tests Problem: The "Slave I/O thread couldn't register on master" error sporadically occurred in replication tests because the slave I/O thread got killed by STOP SLAVE before or while registering on master. Fixed by checking the state of the I/O thread, and issueing the error only if it was not explicitely killed by a user.
2007-11-22 17:22:54 +01:00
if (!check_io_slave_killed(thd, mi, "Slave I/O thread killed "
"while registering slave on master"))
{
sql_print_error("Slave I/O thread couldn't register on master");
if (try_to_reconnect(thd, mysql, mi, &retry_count, suppress_warnings,
reconnect_messages[SLAVE_RECON_ACT_REG]))
goto err;
}
else
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
goto err;
goto connected;
BUG#21123 (Slave fails to reconnect on update_slave_list): Fixing comments and adding error message if slave fails to register on master.
2007-05-31 10:17:16 +02:00
}
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
DBUG_EXECUTE_IF("FORCE_SLAVE_TO_RECONNECT_REG",
if (!retry_count_reg)
{
retry_count_reg++;
sql_print_information("Forcing to reconnect slave I/O thread");
if (try_to_reconnect(thd, mysql, mi, &retry_count, suppress_warnings,
reconnect_messages[SLAVE_RECON_ACT_REG]))
goto err;
goto connected;
});
work to enable reading 3.23 logs - not yet finished moved fail-safe replication routines from sql_repl.cc to repl_failsafe.cc write start event only in the first log
2001-11-11 06:24:12 +01:00
}
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
Use our version of RWLOCKS on UNIXWARE 7 More DBUG info for replication Better error messages from replication Fixed bug in replication code when connecting to 'localhost' (time was not released properly) Block ALARM signal on Linux for signal handler thread (Fixes problem with running mysqld with --debug) Removed warning when setting an AUTO_INCREMENT field to NULL
2002-08-21 21:04:22 +02:00
DBUG_PRINT("info",("Starting reading binary log from master"));
fixed replication assert failure ( found by Jeremy Zawodny)
2002-03-08 23:02:11 +01:00
while (!io_slave_killed(thd,mi))
keep bk happy
2001-01-22 03:46:32 +01:00
{
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Requesting binlog dump");
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
if (request_dump(mysql, mi, &suppress_warnings))
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
{
sql_print_error("Failed on request_dump()");
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
if (check_io_slave_killed(thd, mi, "Slave I/O thread killed while \
requesting master dump") ||
try_to_reconnect(thd, mysql, mi, &retry_count, suppress_warnings,
reconnect_messages[SLAVE_RECON_ACT_DUMP]))
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
goto connected;
}
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
DBUG_EXECUTE_IF("FORCE_SLAVE_TO_RECONNECT_DUMP",
if (!retry_count_dump)
{
retry_count_dump++;
sql_print_information("Forcing to reconnect slave I/O thread");
if (try_to_reconnect(thd, mysql, mi, &retry_count, suppress_warnings,
reconnect_messages[SLAVE_RECON_ACT_DUMP]))
goto err;
goto connected;
});
Import changeset
2000-07-31 21:29:14 +02:00
BUG#44270: RESET SLAVE does not reset Last_IO_Error or Last_IO_Errno The server was not cleaning the last IO error and error number when resetting slave. This patch addresses this issue by backporting into 5.1 part of the patch in BUG 34654. A fix for this issue had already been pushed into 6.0 as part of the aforementioned bug, however the patch also included some refactoring. The fix for 5.1 does not take into account the refactoring part.
2009-06-03 16:14:18 +02:00
DBUG_ASSERT(mi->last_error().number == 0);
fixed replication assert failure ( found by Jeremy Zawodny)
2002-03-08 23:02:11 +01:00
while (!io_slave_killed(thd,mi))
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
{
Fixed compiler warnings detected by option -Wshadow and -Wunused: - Removed not used variables and functions - Added #ifdef around code that is not used - Renamed variables and functions to avoid conflicts - Removed some not used arguments Fixed some class/struct warnings in ndb Added define IS_LONGDATA() to simplify code in libmysql.c I did run gcov on the changes and added 'purecov' comments on almost all lines that was not just variable name changes
2006-12-14 23:51:37 +01:00
ulong event_len;
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
/*
Clearer states in SHOW PROCESSLIST for replication threads. For example the Binlog_dump thread (on the master) sometimes showed "Slave:". And there were confusing messages where "binlog" was employed instead of "relay log".
2003-08-25 14:13:58 +02:00
We say "waiting" because read_event() will wait if there's nothing to
auto-ROLLBACK if binlog was not closed properly auto-commit on Xid_log_event
2005-02-09 20:04:28 +01:00
read. But if there's something to read, it will not wait. The
important thing is to not confuse users by saying "reading" whereas
we're in fact receiving nothing.
Clearer states in SHOW PROCESSLIST for replication threads. For example the Binlog_dump thread (on the master) sometimes showed "Slave:". And there were confusing messages where "binlog" was employed instead of "relay log".
2003-08-25 14:13:58 +02:00
*/
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Waiting for master to send event");
Fixed compiler warnings detected by option -Wshadow and -Wunused: - Removed not used variables and functions - Added #ifdef around code that is not used - Renamed variables and functions to avoid conflicts - Removed some not used arguments Fixed some class/struct warnings in ndb Added define IS_LONGDATA() to simplify code in libmysql.c I did run gcov on the changes and added 'purecov' comments on almost all lines that was not just variable name changes
2006-12-14 23:51:37 +01:00
event_len= read_event(mysql, mi, &suppress_warnings);
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
if (check_io_slave_killed(thd, mi, "Slave I/O thread killed while \
reading event"))
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
DBUG_EXECUTE_IF("FORCE_SLAVE_TO_RECONNECT_EVENT",
if (!retry_count_event)
{
retry_count_event++;
sql_print_information("Forcing to reconnect slave I/O thread");
if (try_to_reconnect(thd, mysql, mi, &retry_count, suppress_warnings,
reconnect_messages[SLAVE_RECON_ACT_EVENT]))
goto err;
goto connected;
});
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
if (event_len == packet_error)
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
uint mysql_error_number= mysql_errno(mysql);
Fix for bug #24192 "MySQL replication does not exit server when running out of memory" In case of out-of-memory error received from the master, print the corresponding message to the error log and stop slave I/O thread to avoid reconnecting with a wrong binary log position.
2007-07-11 16:38:45 +02:00
switch (mysql_error_number) {
case CR_NET_PACKET_TOO_LARGE:
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
sql_print_error("\
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Log entry on master is longer than max_allowed_packet (%ld) on \
slave. If the entry is correct, restart the server with a higher value of \
max_allowed_packet",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
thd->variables.max_allowed_packet);
Bug #42914 Log event that larger than max_allowed_packet results in stop of slave I/O thread, But there is no Last_IO_Error reported. On the master, if a binary log event is larger than max_allowed_packet, ER_MASTER_FATAL_ERROR_READING_BINLOG and the specific reason of this error is sent to a slave when it requests a dump from the master, thus leading the I/O thread to stop. On a slave, the I/O thread stops when receiving a packet larger than max_allowed_packet. In both cases, however, there was no Last_IO_Error reported. This patch adds code to report the Last_IO_Error and exact reason before stopping the I/O thread and also reports the case the out memory pops up while handling packets from the master.
2009-09-18 10:20:29 +02:00
mi->report(ERROR_LEVEL, ER_NET_PACKET_TOO_LARGE,
Bug #43414 Parenthesis (and other) warnings compiling MySQL with gcc 4.3.2 Cleaning up warnings not present in 5.0.
2009-09-23 15:21:29 +02:00
"%s", ER(ER_NET_PACKET_TOO_LARGE));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
Fix for bug #24192 "MySQL replication does not exit server when running out of memory" In case of out-of-memory error received from the master, print the corresponding message to the error log and stop slave I/O thread to avoid reconnecting with a wrong binary log position.
2007-07-11 16:38:45 +02:00
case ER_MASTER_FATAL_ERROR_READING_BINLOG:
Bug #42914 Log event that larger than max_allowed_packet results in stop of slave I/O thread, But there is no Last_IO_Error reported. On the master, if a binary log event is larger than max_allowed_packet, ER_MASTER_FATAL_ERROR_READING_BINLOG and the specific reason of this error is sent to a slave when it requests a dump from the master, thus leading the I/O thread to stop. On a slave, the I/O thread stops when receiving a packet larger than max_allowed_packet. In both cases, however, there was no Last_IO_Error reported. This patch adds code to report the Last_IO_Error and exact reason before stopping the I/O thread and also reports the case the out memory pops up while handling packets from the master.
2009-09-18 10:20:29 +02:00
mi->report(ERROR_LEVEL, ER_MASTER_FATAL_ERROR_READING_BINLOG,
ER(ER_MASTER_FATAL_ERROR_READING_BINLOG),
mysql_error_number, mysql_error(mysql));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
Bug #42914 Log event that larger than max_allowed_packet results in stop of slave I/O thread, But there is no Last_IO_Error reported. On the master, if a binary log event is larger than max_allowed_packet, ER_MASTER_FATAL_ERROR_READING_BINLOG and the specific reason of this error is sent to a slave when it requests a dump from the master, thus leading the I/O thread to stop. On a slave, the I/O thread stops when receiving a packet larger than max_allowed_packet. In both cases, however, there was no Last_IO_Error reported. This patch adds code to report the Last_IO_Error and exact reason before stopping the I/O thread and also reports the case the out memory pops up while handling packets from the master.
2009-09-18 10:20:29 +02:00
case ER_OUT_OF_RESOURCES:
Fix for bug #24192 "MySQL replication does not exit server when running out of memory" In case of out-of-memory error received from the master, print the corresponding message to the error log and stop slave I/O thread to avoid reconnecting with a wrong binary log position.
2007-07-11 16:38:45 +02:00
sql_print_error("\
Stopping slave I/O thread due to out-of-memory error from master");
Bug #42914 Log event that larger than max_allowed_packet results in stop of slave I/O thread, But there is no Last_IO_Error reported. On the master, if a binary log event is larger than max_allowed_packet, ER_MASTER_FATAL_ERROR_READING_BINLOG and the specific reason of this error is sent to a slave when it requests a dump from the master, thus leading the I/O thread to stop. On a slave, the I/O thread stops when receiving a packet larger than max_allowed_packet. In both cases, however, there was no Last_IO_Error reported. This patch adds code to report the Last_IO_Error and exact reason before stopping the I/O thread and also reports the case the out memory pops up while handling packets from the master.
2009-09-18 10:20:29 +02:00
mi->report(ERROR_LEVEL, ER_OUT_OF_RESOURCES,
Bug #43414 Parenthesis (and other) warnings compiling MySQL with gcc 4.3.2 Cleaning up warnings not present in 5.0.
2009-09-23 15:21:29 +02:00
"%s", ER(ER_OUT_OF_RESOURCES));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
}
Fix for bug #19328: Slave timeout with COM_REGISTER_SLAVE error causing stop Problem: "Under high load, the slave registering to the master can timeout during the COM_REGISTER_SLAVE execution. This causes an error, which prevents the slave from connecting at all." Fix: Do not abort immediately, but retry registering on master.
2007-06-29 18:48:22 +02:00
if (try_to_reconnect(thd, mysql, mi, &retry_count, suppress_warnings,
reconnect_messages[SLAVE_RECON_ACT_EVENT]))
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
goto connected;
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
} // if (event_len == packet_error)
post-review fixes. Now ROLLBACK is done in Format_description_log_event
2005-02-17 13:52:16 +01:00
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
retry_count=0; // ok event, reset retry counter
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Queueing master event to the relay log");
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
if (queue_event(mi,(const char*)mysql->net.read_pos + 1,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
event_len))
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
{
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
mi->report(ERROR_LEVEL, ER_SLAVE_RELAY_LOG_WRITE_FAILURE,
ER(ER_SLAVE_RELAY_LOG_WRITE_FAILURE),
"could not queue event from master");
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
}
many warnings (practically safe but annoying) corrected
2006-01-03 17:54:54 +01:00
if (flush_master_info(mi, 1))
{
sql_print_error("Failed to flush master info file");
goto err;
}
- Fix for memory leak if the SQL slave thread is killed just after reading an event. - A few more mutex locks and unlocks of rli.log_space_lock for doing clean reads of rli.ignore_log_space_limit - Broadcast after unlock, not before (small speed optimisation).
2003-05-25 23:09:46 +02:00
/*
See if the relay logs take too much space.
We don't lock mi->rli.log_space_lock here; this dirty read saves time
and does not introduce any problem:
- if mi->rli.ignore_log_space_limit is 1 but becomes 0 just after (so
the clean value is 0), then we are reading only one more event as we
should, and we'll block only at the next event. No big deal.
- if mi->rli.ignore_log_space_limit is 0 but becomes 1 just after (so
the clean value is 1), then we are going into wait_for_relay_log_space()
for no reason, but this function will do a clean read, notice the clean
value and exit immediately.
*/
Fix for nightly build test failure (test update). More messages. Testcase for bug 651.
2003-06-16 15:49:54 +02:00
#ifndef DBUG_OFF
{
char llbuf1[22], llbuf2[22];
DBUG_PRINT("info", ("log_space_limit=%s log_space_total=%s \
ignore_log_space_limit=%d",
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
llstr(rli->log_space_limit,llbuf1),
llstr(rli->log_space_total,llbuf2),
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
(int) rli->ignore_log_space_limit));
Fix for nightly build test failure (test update). More messages. Testcase for bug 651.
2003-06-16 15:49:54 +02:00
}
#endif
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
if (rli->log_space_limit && rli->log_space_limit <
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
rli->log_space_total &&
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
!rli->ignore_log_space_limit)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
if (wait_for_relay_log_space(rli))
{
sql_print_error("Slave I/O thread aborted while waiting for relay \
relay_log_space_limit DBUG_ macro cleanup buffer boundary cleanup This changeset, although not fully tested, works for me better than anything I've had so far, including what is in the repository. I will push it unless something crashes while I am writing this :-)
2002-04-02 06:46:23 +02:00
log space");
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
}
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
}
Import changeset
2000-07-31 21:29:14 +02:00
Portability fixes
2001-04-20 14:18:46 +02:00
// error = 0;
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
err:
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
// print the current replication position
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
sql_print_information("Slave I/O thread exiting, read up to log '%s', position %s",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
IO_RPL_LOG_NAME, llstr(mi->master_log_pos,llbuff));
Bug #38816: kill + flush tables with read lock + stored procedures causes crashes! The problem of that bugreport was mostly fixed by the patch for bug 38691. However, attached test case focused on another crash or valgrind warning problem: SHOW PROCESSLIST query accesses freed memory of SP instruction that run in a parallel connection. Changes of thd->query/thd->query_length in dangerous places have been guarded with the per-thread LOCK_thd_data mutex (the THD::LOCK_delete mutex has been renamed to THD::LOCK_thd_data).
2009-07-24 17:58:58 +02:00
thd->set_query(NULL, 0);
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
thd->reset_db(NULL, 0);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
if (mysql)
{
Bug#19437 (Connection refused by server: "2002 Can't connect... /master.sock"): Clearing active VIO before calling mysql_close() in the slave I/O thread.
2006-06-20 20:46:45 +02:00
/*
Here we need to clear the active VIO before closing the
connection with the master. The reason is that THD::awake()
might be called from terminate_slave_thread() because somebody
issued a STOP SLAVE. If that happends, the close_active_vio()
can be called in the middle of closing the VIO associated with
the 'mysql' object, causing a crash.
*/
#ifdef SIGNAL_WITH_VIO_CLOSE
thd->clear_active_vio();
#endif
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
mysql_close(mysql);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
mi->mysql=0;
}
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
write_ignored_events_info_to_relay_log(thd, mi);
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Waiting for slave mutex on exit");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_mutex_lock(&mi->run_lock);
Fixes bug #13377. Added code to close active log files in case of log reset condition.
2005-10-03 18:34:42 +02:00
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
/* Forget the relay log's format */
delete mi->rli.relay_log.description_event_for_queue;
mi->rli.relay_log.description_event_for_queue= 0;
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
// TODO: make rpl_status part of Master_info
fixes for mysql-test to run without manager a bit more work on failsafe replication
2001-10-11 21:54:06 +02:00
change_rpl_status(RPL_ACTIVE_SLAVE,RPL_IDLE_SLAVE);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(thd->net.buff != 0);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
net_end(&thd->net); // destructor will not free it, because net.vio is 0
Bug#30882 Dropping a temporary table inside a stored function may cause a server crash If a stored function that contains a drop temporary table statement is invoked by a create temporary table of the same name may cause a server crash. The problem is that when dropping a table no check is done to ensure that table is not being used by some outer query (or outer statement), potentially leaving the outer query with a reference to a stale (freed) table. The solution is when dropping a temporary table, always check if the table is being used by some outer statement as a temporary table can be dropped inside stored procedures. The check is performed by looking at the TABLE::query_id value for temporary tables. To simplify this check and to solve a bug related to handling of temporary tables in prelocked mode, this patch changes the way in which this member is used to track the fact that table is used/unused. Now we ensure that TABLE::query_id is zero for unused temporary tables (which means that all temporary tables which were used by a statement should be marked as free for reuse after it's execution has been completed).
2007-11-01 21:52:56 +01:00
close_thread_tables(thd);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_mutex_lock(&LOCK_thread_count);
replication updates. This changeset seems to be working fine on test systems. If no problems are discovered in the next week, this will make the replication code ready for 4.0.2 release.
2002-03-30 20:36:05 +01:00
THD_CHECK_SENTRY(thd);
Import changeset
2000-07-31 21:29:14 +02:00
delete thd;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_mutex_unlock(&LOCK_thread_count);
slave.cc: BUG#20850: Assert during slave shutdown in many rpl_* tests This patch has already been applied to 5.1.
2006-07-25 12:30:18 +02:00
mi->abort_slave= 0;
mi->slave_running= 0;
mi->io_thd= 0;
BUG#25306 (Race conditions during replication slave shutdown (valgrind stacks)): The possibility of the race is removed by changing sequence of calls pthread_mutex_unlock(&mi->run_lock); pthread_cond_broadcast(&mi->stop_cond); into pthread_cond_broadcast(&mi->stop_cond); pthread_mutex_unlock(&mi->run_lock); at the end of I/O thread (similar change at the end of SQL thread). This ensures that no thread waiting on the condition executes between the broadcast and the unlock and thus can't delete the mi structure which caused the bug.
2007-02-03 20:14:16 +01:00
/*
Note: the order of the two following calls (first broadcast, then unlock)
is important. Otherwise a killer_thread can execute between the calls and
delete the mi structure leading to a crash! (see BUG#25306 for details)
*/
Bug#20850 (Assert during slave shutdown in many rpl_* tests): Previous patch didn't work: moving variable settings inside critical region.
2006-07-07 18:31:00 +02:00
pthread_cond_broadcast(&mi->stop_cond); // tell the world we are done
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
DBUG_EXECUTE_IF("simulate_slave_delay_at_terminate_bug38694", sleep(5););
BUG#25306 (Race conditions during replication slave shutdown (valgrind stacks)): The possibility of the race is removed by changing sequence of calls pthread_mutex_unlock(&mi->run_lock); pthread_cond_broadcast(&mi->stop_cond); into pthread_cond_broadcast(&mi->stop_cond); pthread_mutex_unlock(&mi->run_lock); at the end of I/O thread (similar change at the end of SQL thread). This ensures that no thread waiting on the condition executes between the broadcast and the unlock and thus can't delete the mi structure which caused the bug.
2007-02-03 20:14:16 +01:00
pthread_mutex_unlock(&mi->run_lock);
A patch for Bug#47474 (mysqld hits Dbug_violation_helper assert when compiled with Sun Studio compiler). The thing is that Sun Studio compiler calls destructor of stack objects when pthread_exit() is called. That triggered an assertion in DBUG_ENTER()/DBUG_RETURN() validation logic (if DBUG_ENTER() is used in the beginning of function, all returns should be replaced by DBUG_RETURN/DBUG_VOID_RETURN macros). A fix is to explicitly use DBUG_LEAVE macro.
2009-09-23 15:10:23 +02:00
DBUG_LEAVE; // Must match DBUG_ENTER()
Bug#20850 (Assert during slave shutdown in many rpl_* tests): Previous patch didn't work: moving variable settings inside critical region.
2006-07-07 18:31:00 +02:00
my_thread_end();
Import changeset
2000-07-31 21:29:14 +02:00
pthread_exit(0);
A patch for Bug#47474 (mysqld hits Dbug_violation_helper assert when compiled with Sun Studio compiler). The thing is that Sun Studio compiler calls destructor of stack objects when pthread_exit() is called. That triggered an assertion in DBUG_ENTER()/DBUG_RETURN() validation logic (if DBUG_ENTER() is used in the beginning of function, all returns should be replaced by DBUG_RETURN/DBUG_VOID_RETURN macros). A fix is to explicitly use DBUG_LEAVE macro.
2009-09-23 15:10:23 +02:00
return 0; // Avoid compiler warnings
Import changeset
2000-07-31 21:29:14 +02:00
}
Bug #42861 Assigning invalid directories to --slave-load-tmpdir crashes the slave Compiling with debug and assigning an invalid directory to --slave-load-tmpdir was crashing the slave due to the following assertion DBUG_ASSERT(! is_set() || can_overwrite_status). This assertion assumes that a thread can change its state once (i.e. ok,error, etc) before aborting, cleaning/resuming or completing its execution unless the overwrite flag (i.e. can_overwrite_status) is true. The Append_block_log_event::do_apply_event which is responsible for creating temporary file(s) was not cleaning the thread state. Thus a failure while trying to create a file in an invalid temporary directory was causing the crash. To fix the problem we check if the temporary directory is valid before starting the SQL Thread and reset the thread state before creating a file in Append_block_log_event::do_apply_event.
2009-03-18 11:31:17 +01:00
/*
Check the temporary directory used by commands like
LOAD DATA INFILE.
*/
static
BUG#43949 Initialization of slave produces a warning message in Valgrind In order to define the --slave-load-tmpdir, the init_relay_log_file() was calling fn_format(MY_PACK_FILENAME) which internally was indirectly calling strmov_overlapp() (through pack_dirname) and the following warning message was being printed out while running in Valgrind: "source and destination overlap in strcpy". We fixed the issue by removing the flag MY_PACK_FILENAME as it was not necessary. In a nutshell, with this flag the function fn_format() tried to replace a directory by either "~", "." or "..". However, we wanted exactly to remove such strings. In this patch, we also refactored the functions init_relay_log_file() and check_temp_dir(). The former was refactored to call the fn_format() with the flag MY_SAFE_PATH along with the MY_RETURN_REAL_PATH, in order to avoid issues with long directories and return an absolute path, respectively. The flag MY_SAFE_UNPACK_FILENAME was removed too as it was responsible for removing "~", "." or ".." only from the file parameter and we wanted to remove such strings from the directory parameter in the fn_format(). This result is stored in an rli variable, which is then processed by the other function in order to verify if the directory exists and if we are able to create files in it.
2009-04-19 03:21:33 +02:00
int check_temp_dir(char* tmp_file)
Bug #42861 Assigning invalid directories to --slave-load-tmpdir crashes the slave Compiling with debug and assigning an invalid directory to --slave-load-tmpdir was crashing the slave due to the following assertion DBUG_ASSERT(! is_set() || can_overwrite_status). This assertion assumes that a thread can change its state once (i.e. ok,error, etc) before aborting, cleaning/resuming or completing its execution unless the overwrite flag (i.e. can_overwrite_status) is true. The Append_block_log_event::do_apply_event which is responsible for creating temporary file(s) was not cleaning the thread state. Thus a failure while trying to create a file in an invalid temporary directory was causing the crash. To fix the problem we check if the temporary directory is valid before starting the SQL Thread and reset the thread state before creating a file in Append_block_log_event::do_apply_event.
2009-03-18 11:31:17 +01:00
{
int fd;
MY_DIR *dirp;
BUG#43949 Initialization of slave produces a warning message in Valgrind In order to define the --slave-load-tmpdir, the init_relay_log_file() was calling fn_format(MY_PACK_FILENAME) which internally was indirectly calling strmov_overlapp() (through pack_dirname) and the following warning message was being printed out while running in Valgrind: "source and destination overlap in strcpy". We fixed the issue by removing the flag MY_PACK_FILENAME as it was not necessary. In a nutshell, with this flag the function fn_format() tried to replace a directory by either "~", "." or "..". However, we wanted exactly to remove such strings. In this patch, we also refactored the functions init_relay_log_file() and check_temp_dir(). The former was refactored to call the fn_format() with the flag MY_SAFE_PATH along with the MY_RETURN_REAL_PATH, in order to avoid issues with long directories and return an absolute path, respectively. The flag MY_SAFE_UNPACK_FILENAME was removed too as it was responsible for removing "~", "." or ".." only from the file parameter and we wanted to remove such strings from the directory parameter in the fn_format(). This result is stored in an rli variable, which is then processed by the other function in order to verify if the directory exists and if we are able to create files in it.
2009-04-19 03:21:33 +02:00
char tmp_dir[FN_REFLEN];
size_t tmp_dir_size;
Bug #42861 Assigning invalid directories to --slave-load-tmpdir crashes the slave Compiling with debug and assigning an invalid directory to --slave-load-tmpdir was crashing the slave due to the following assertion DBUG_ASSERT(! is_set() || can_overwrite_status). This assertion assumes that a thread can change its state once (i.e. ok,error, etc) before aborting, cleaning/resuming or completing its execution unless the overwrite flag (i.e. can_overwrite_status) is true. The Append_block_log_event::do_apply_event which is responsible for creating temporary file(s) was not cleaning the thread state. Thus a failure while trying to create a file in an invalid temporary directory was causing the crash. To fix the problem we check if the temporary directory is valid before starting the SQL Thread and reset the thread state before creating a file in Append_block_log_event::do_apply_event.
2009-03-18 11:31:17 +01:00
DBUG_ENTER("check_temp_dir");
BUG#43949 Initialization of slave produces a warning message in Valgrind In order to define the --slave-load-tmpdir, the init_relay_log_file() was calling fn_format(MY_PACK_FILENAME) which internally was indirectly calling strmov_overlapp() (through pack_dirname) and the following warning message was being printed out while running in Valgrind: "source and destination overlap in strcpy". We fixed the issue by removing the flag MY_PACK_FILENAME as it was not necessary. In a nutshell, with this flag the function fn_format() tried to replace a directory by either "~", "." or "..". However, we wanted exactly to remove such strings. In this patch, we also refactored the functions init_relay_log_file() and check_temp_dir(). The former was refactored to call the fn_format() with the flag MY_SAFE_PATH along with the MY_RETURN_REAL_PATH, in order to avoid issues with long directories and return an absolute path, respectively. The flag MY_SAFE_UNPACK_FILENAME was removed too as it was responsible for removing "~", "." or ".." only from the file parameter and we wanted to remove such strings from the directory parameter in the fn_format(). This result is stored in an rli variable, which is then processed by the other function in order to verify if the directory exists and if we are able to create files in it.
2009-04-19 03:21:33 +02:00
/*
Get the directory from the temporary file.
*/
dirname_part(tmp_dir, tmp_file, &tmp_dir_size);
Bug #42861 Assigning invalid directories to --slave-load-tmpdir crashes the slave Compiling with debug and assigning an invalid directory to --slave-load-tmpdir was crashing the slave due to the following assertion DBUG_ASSERT(! is_set() || can_overwrite_status). This assertion assumes that a thread can change its state once (i.e. ok,error, etc) before aborting, cleaning/resuming or completing its execution unless the overwrite flag (i.e. can_overwrite_status) is true. The Append_block_log_event::do_apply_event which is responsible for creating temporary file(s) was not cleaning the thread state. Thus a failure while trying to create a file in an invalid temporary directory was causing the crash. To fix the problem we check if the temporary directory is valid before starting the SQL Thread and reset the thread state before creating a file in Append_block_log_event::do_apply_event.
2009-03-18 11:31:17 +01:00
/*
Check if the directory exists.
*/
if (!(dirp=my_dir(tmp_dir,MYF(MY_WME))))
DBUG_RETURN(1);
my_dirend(dirp);
/*
Check permissions to create a file.
*/
if ((fd= my_create(tmp_file, CREATE_MODE,
O_WRONLY | O_BINARY | O_EXCL | O_NOFOLLOW,
MYF(MY_WME))) < 0)
DBUG_RETURN(1);
/*
Clean up.
*/
my_close(fd, MYF(0));
my_delete(tmp_file, MYF(0));
DBUG_RETURN(0);
}
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
BUG#41924: high-level replication functions are not commented Adding comments to some of the high-level functions in replication.
2009-01-09 13:49:24 +01:00
/**
Slave SQL thread entry point.
@param arg Pointer to Relay_log_info object that holds information
for the SQL thread.
keep bk happy
2001-01-22 03:46:32 +01:00
BUG#41924: high-level replication functions are not commented Adding comments to some of the high-level functions in replication.
2009-01-09 13:49:24 +01:00
@return Always 0.
*/
pthread_handler_decl() changed to be ctags-friendly (and contain extern "C" when necessary)
2005-10-08 16:39:55 +02:00
pthread_handler_t handle_slave_sql(void *arg)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
THD *thd; /* needs to be first for thread_stack */
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
char llbuff[22],llbuff1[22];
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
Relay_log_info* rli = &((Master_info*)arg)->rli;
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
const char *errmsg;
// needs to call my_thread_init(), otherwise we get a coredump in DBUG_ stuff
my_thread_init();
Fix that round(0.1,1) == round(0.1,1)
2003-03-10 11:00:19 +01:00
DBUG_ENTER("handle_slave_sql");
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(rli->inited);
pthread_mutex_lock(&rli->run_lock);
DBUG_ASSERT(!rli->slave_running);
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
errmsg= 0;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
#ifndef DBUG_OFF
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
rli->events_till_abort = abort_slave_event_count;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
#endif
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
thd = new THD; // note that contructor of THD uses DBUG_ !
Added support for ULONG division with DIV Fixed non fatal memory leak in slave code.
2003-03-19 20:23:13 +01:00
thd->thread_stack = (char*)&thd; // remember where our stack is
Bug #33931 assertion at write_ignored_events_info_to_relay_log if init_slave_thread() fails and bug#33932 assertion at handle_slave_sql if init_slave_thread() fails the asserts were caused by bug33931: having thd deleted at time of executing err: code plus a missed initialization; bug33932: initialization of slave_is_running member was missed; fixed with relocating mi members initialization and removing delete thd It is safe to do as deletion happens later explicitly in the caller of init_slave_thread(). Todo: at merging the test is better to be moved into suite/bugs for 5.x (when x>0).
2008-02-13 13:09:41 +01:00
rli->sql_thd= thd;
Added support for ULONG division with DIV Fixed non fatal memory leak in slave code.
2003-03-19 20:23:13 +01:00
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
/* Inform waiting threads that slave has started */
rli->slave_run_id++;
Bug #33931 assertion at write_ignored_events_info_to_relay_log if init_slave_thread() fails and bug#33932 assertion at handle_slave_sql if init_slave_thread() fails the asserts were caused by bug33931: having thd deleted at time of executing err: code plus a missed initialization; bug33932: initialization of slave_is_running member was missed; fixed with relocating mi members initialization and removing delete thd It is safe to do as deletion happens later explicitly in the caller of init_slave_thread(). Todo: at merging the test is better to be moved into suite/bugs for 5.x (when x>0).
2008-02-13 13:09:41 +01:00
rli->slave_running = 1;
Give better error from reconnect() Fixed hang in start_slave_threads() when thread dies quickly.
2002-08-24 04:44:16 +02:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_detach_this_thread();
if (init_slave_thread(thd, SLAVE_THD_SQL))
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
{
/*
TODO: this is currently broken - slave start and change master
will be stuck if we fail here
*/
pthread_cond_broadcast(&rli->start_cond);
pthread_mutex_unlock(&rli->run_lock);
BUG#38197 Errors in @@init_slave not visible in 'show slave status' Some errors that cause the slave SQL thread to stop are not shown in the Slave_SQL_Error column of "SHOW SLAVE STATUS". Instead, the error is only in the server's error log. That makes it difficult to analyze the error for the user. One example of an error that stops the slave but is not shown by "SHOW SLAVE STATUS" is when @@global.init_slave is set incorrectly (e.g., it contains something that is not valid SQL). Three failures were not correctly reported: 1 - Failures during slave thread initialization 2 - Failures while initializing the relay log position right after starting the slave thread. 3 - Failures while processing queries passed through the init_slave option. This patch fixes the issues by reporting the errors through relay-info->report.
2009-02-11 12:56:25 +01:00
rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
"Failed during slave thread initialization");
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
goto err;
}
Added support for ULONG division with DIV Fixed non fatal memory leak in slave code.
2003-03-19 20:23:13 +01:00
thd->init_for_queries();
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
thd->temporary_tables = rli->save_temporary_tables; // restore temp tables
BUG#41725: slave crashes when inserting into temporary table after stop/start slave When stopping and restarting the slave while it is replicating temporary tables, the server would crash or raise an assertion failure. This was due to the fact that although temporary tables are saved between slave threads restart, the reference to the thread in use (table->in_use) was not being properly updated when the restart happened (it would still reference the old/invalid thread instead of the new one). This patch addresses this issue by resetting the reference to the new slave thread on slave thread restart.
2009-05-23 01:15:21 +02:00
set_thd_in_use_temporary_tables(rli); // (re)set sql_thd in use for saved temp tables
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
pthread_mutex_lock(&LOCK_thread_count);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
threads.append(thd);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
pthread_mutex_unlock(&LOCK_thread_count);
A change of behaviour of Seconds_Behind_Master from SHOW SLAVE STATUS. It's going into 4.1 because old behaviour was somewhat nonsensical (kind of bug). Changes are that if repl threads are down or disconnected the column will be NULL, and if master is idle the column will not grow indefinitely anymore.
2004-12-16 18:12:22 +01:00
/*
We are going to set slave_running to 1. Assuming slave I/O thread is
alive and connected, this is going to make Seconds_Behind_Master be 0
i.e. "caught up". Even if we're just at start of thread. Well it's ok, at
the moment we start we can think we are caught up, and the next second we
start receiving data so we realize we are not caught up and
Seconds_Behind_Master grows. No big deal.
*/
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
rli->abort_slave = 0;
pthread_mutex_unlock(&rli->run_lock);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
pthread_cond_broadcast(&rli->start_cond);
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
2 bugfixes: - Bug #985: "Between RESET SLAVE and START SLAVE, SHOW SLAVE STATUS is wrong." Now RESET SLAVE puts correct info in mi->host etc. A new test rpl_reset_slave for that. - Bug #986: "CHANGE MASTER & START SLAVE do not reset error columns in SHOW SLAVE STATUS". Now these reset the errors.
2003-08-04 10:59:44 +02:00
/*
Reset errors for a clean start (otherwise, if the master is idle, the SQL
thread may execute no Query_log_event, so the error will remain even
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
though there's no problem anymore). Do not reset the master timestamp
Don't flush cur_log (relay log) on flush_relay_log_info becasue this crashes the server if cur_log is 'hot' and the io_thread has changed log file. Updated project files for windows Made rpl_change_master.test portable Ensure that mutex are not freed if not initilized
2003-11-22 02:21:40 +01:00
(imagine the slave has caught everything, the STOP SLAVE and START SLAVE:
as we are not sure that we are going to receive a query, we want to
remember the last master timestamp (to say how many seconds behind we are
now.
Final push for WL#1098: "Add a column "Timestamp_of_last_master_event_executed" in SHOW SLAVE STATUS". Finally this is adding - Slave_IO_State (a copy of the State column of SHOW PROCESSLIST for the I/O thread, so that the users, most of the time, has enough info with only SHOW SLAVE STATUS). - Seconds_behind_master. When the slave connects to the master it does SELECT UNIX_TIMESTAMP() on the master, computes the absolute difference between the master's and the slave's clock. It records the timestamp of the last event executed by the SQL thread, and does a small computation to find the number of seconds by which the slave is late.
2003-10-09 00:06:21 +02:00
But the master timestamp is reset by RESET SLAVE & CHANGE MASTER.
2 bugfixes: - Bug #985: "Between RESET SLAVE and START SLAVE, SHOW SLAVE STATUS is wrong." Now RESET SLAVE puts correct info in mi->host etc. A new test rpl_reset_slave for that. - Bug #986: "CHANGE MASTER & START SLAVE do not reset error columns in SHOW SLAVE STATUS". Now these reset the errors.
2003-08-04 10:59:44 +02:00
*/
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
rli->clear_error();
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
//tell the I/O thread to take relay_log_space_limit into account from now on
- Fix for memory leak if the SQL slave thread is killed just after reading an event. - A few more mutex locks and unlocks of rli.log_space_lock for doing clean reads of rli.ignore_log_space_limit - Broadcast after unlock, not before (small speed optimisation).
2003-05-25 23:09:46 +02:00
pthread_mutex_lock(&rli->log_space_lock);
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
rli->ignore_log_space_limit= 0;
- Fix for memory leak if the SQL slave thread is killed just after reading an event. - A few more mutex locks and unlocks of rli.log_space_lock for doing clean reads of rli.ignore_log_space_limit - Broadcast after unlock, not before (small speed optimisation).
2003-05-25 23:09:46 +02:00
pthread_mutex_unlock(&rli->log_space_lock);
"After Monty's review" changes to the fix for BUG#8325 "Deadlock in replication thread stops replication": s/sleep/safe_sleep (thread safe); sleep 0/1/2/3/4/5/5/5 (get slave less late); no message on error log (deadlock is too common sometimes), a global counter instead (SHOW STATUS LIKE 'slave_retried_transactions'). Plus a fix for libmysql/Makefile.shared
2005-03-23 19:19:36 +01:00
rli->trans_retries= 0; // start from "no error"
Manual merge 5.0-rpl -> 5.1-rpl. involved bug#12691, bug#27571
2007-10-30 12:49:42 +01:00
DBUG_PRINT("info", ("rli->trans_retries: %lu", rli->trans_retries));
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
if (init_relay_log_pos(rli,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
rli->group_relay_log_name,
rli->group_relay_log_pos,
1 /*need data lock*/, &errmsg,
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
1 /*look for a description_event*/))
BUG#38197 Errors in @@init_slave not visible in 'show slave status' Some errors that cause the slave SQL thread to stop are not shown in the Slave_SQL_Error column of "SHOW SLAVE STATUS". Instead, the error is only in the server's error log. That makes it difficult to analyze the error for the user. One example of an error that stops the slave but is not shown by "SHOW SLAVE STATUS" is when @@global.init_slave is set incorrectly (e.g., it contains something that is not valid SQL). Three failures were not correctly reported: 1 - Failures during slave thread initialization 2 - Failures while initializing the relay log position right after starting the slave thread. 3 - Failures while processing queries passed through the init_slave option. This patch fixes the issues by reporting the errors through relay-info->report.
2009-02-11 12:56:25 +01:00
{
rli->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
"Error initializing relay log position: %s", errmsg);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
goto err;
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
THD_CHECK_SENTRY(thd);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
#ifndef DBUG_OFF
{
char llbuf1[22], llbuf2[22];
DBUG_PRINT("info", ("my_b_tell(rli->cur_log)=%s rli->event_relay_log_pos=%s",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
llstr(my_b_tell(rli->cur_log),llbuf1),
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
llstr(rli->event_relay_log_pos,llbuf2)));
DBUG_ASSERT(rli->event_relay_log_pos >= BIN_LOG_HEADER_SIZE);
/*
Wonder if this is correct. I (Guilhem) wonder if my_b_tell() returns the
correct position when it's called just after my_b_seek() (the questionable
stuff is those "seek is done on next read" comments in the my_b_seek()
source code).
The crude reality is that this assertion randomly fails whereas
replication seems to work fine. And there is no easy explanation why it
fails (as we my_b_seek(rli->event_relay_log_pos) at the very end of
init_relay_log_pos() called above). Maybe the assertion would be
meaningful if we held rli->data_lock between the my_b_seek() and the
DBUG_ASSERT().
*/
#ifdef SHOULD_BE_CHECKED
DBUG_ASSERT(my_b_tell(rli->cur_log) == rli->event_relay_log_pos);
#endif
}
#endif
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(rli->sql_thd == thd);
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
DBUG_PRINT("master_info",("log_file_name: %s position: %s",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
rli->group_master_log_name,
llstr(rli->group_master_log_pos,llbuff)));
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
if (global_system_variables.log_warnings)
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
sql_print_information("Slave SQL thread initialized, starting replication in \
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
log '%s' at position %s, relay log '%s' position: %s", RPL_LOG_NAME,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
llstr(rli->group_master_log_pos,llbuff),rli->group_relay_log_name,
llstr(rli->group_relay_log_pos,llbuff1));
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
BUG#43949 Initialization of slave produces a warning message in Valgrind In order to define the --slave-load-tmpdir, the init_relay_log_file() was calling fn_format(MY_PACK_FILENAME) which internally was indirectly calling strmov_overlapp() (through pack_dirname) and the following warning message was being printed out while running in Valgrind: "source and destination overlap in strcpy". We fixed the issue by removing the flag MY_PACK_FILENAME as it was not necessary. In a nutshell, with this flag the function fn_format() tried to replace a directory by either "~", "." or "..". However, we wanted exactly to remove such strings. In this patch, we also refactored the functions init_relay_log_file() and check_temp_dir(). The former was refactored to call the fn_format() with the flag MY_SAFE_PATH along with the MY_RETURN_REAL_PATH, in order to avoid issues with long directories and return an absolute path, respectively. The flag MY_SAFE_UNPACK_FILENAME was removed too as it was responsible for removing "~", "." or ".." only from the file parameter and we wanted to remove such strings from the directory parameter in the fn_format(). This result is stored in an rli variable, which is then processed by the other function in order to verify if the directory exists and if we are able to create files in it.
2009-04-19 03:21:33 +02:00
if (check_temp_dir(rli->slave_patternload_file))
Bug #42861 Assigning invalid directories to --slave-load-tmpdir crashes the slave Compiling with debug and assigning an invalid directory to --slave-load-tmpdir was crashing the slave due to the following assertion DBUG_ASSERT(! is_set() || can_overwrite_status). This assertion assumes that a thread can change its state once (i.e. ok,error, etc) before aborting, cleaning/resuming or completing its execution unless the overwrite flag (i.e. can_overwrite_status) is true. The Append_block_log_event::do_apply_event which is responsible for creating temporary file(s) was not cleaning the thread state. Thus a failure while trying to create a file in an invalid temporary directory was causing the crash. To fix the problem we check if the temporary directory is valid before starting the SQL Thread and reset the thread state before creating a file in Append_block_log_event::do_apply_event.
2009-03-18 11:31:17 +01:00
{
rli->report(ERROR_LEVEL, thd->main_da.sql_errno(),
"Unable to use slave's temporary directory %s - %s",
slave_load_tmpdir, thd->main_da.message());
goto err;
}
SCRUM: Task 499 'init_connect, init_slave options'
2003-07-18 11:11:01 +02:00
/* execute init_slave variable */
sys_var_str.value is never NULL
2004-03-20 15:56:20 +01:00
if (sys_init_slave.value_length)
SCRUM: Task 499 'init_connect, init_slave options'
2003-07-18 11:11:01 +02:00
{
Task ID 499:Add a new settable string variable(init_connect, init_slave) to mysqld that is executed for all new connections. (Similar to the client command: mysql_options(... MYSQL_INIT_COMMAND ...).
2003-12-08 06:13:14 +01:00
execute_init_command(thd, &sys_init_slave, &LOCK_sys_init_slave);
Rename: query_error -> is_slave_error. Add comments.
2007-10-19 23:20:38 +02:00
if (thd->is_slave_error)
SCRUM: Task 499 'init_connect, init_slave options'
2003-07-18 11:11:01 +02:00
{
BUG#38197 Errors in @@init_slave not visible in 'show slave status' Some errors that cause the slave SQL thread to stop are not shown in the Slave_SQL_Error column of "SHOW SLAVE STATUS". Instead, the error is only in the server's error log. That makes it difficult to analyze the error for the user. One example of an error that stops the slave but is not shown by "SHOW SLAVE STATUS" is when @@global.init_slave is set incorrectly (e.g., it contains something that is not valid SQL). Three failures were not correctly reported: 1 - Failures during slave thread initialization 2 - Failures while initializing the relay log position right after starting the slave thread. 3 - Failures while processing queries passed through the init_slave option. This patch fixes the issues by reporting the errors through relay-info->report.
2009-02-11 12:56:25 +01:00
rli->report(ERROR_LEVEL, thd->main_da.sql_errno(),
"Slave SQL thread aborted. Can't execute init_slave query");
SCRUM: Task 499 'init_connect, init_slave options'
2003-07-18 11:11:01 +02:00
goto err;
}
}
BUG#13861 - START SLAVE UNTIL may stop 1 evnt too late if log-slave-updates and circul repl Slave SQL thread may execute one extra event when there are events skipped by slave I/O thread (e.g. originated by the same server). Whereas it was requested not to do so by the UNTIL condition. This happens because we compare with the end position of previously executed event. This is fine when there are no skipped by slave I/O thread events, as end position of previous event equals to start position of to be executed event. Otherwise this position equals to start position of skipped event. This is fixed by: - reading the event to be executed before checking if the until condition is satisfied. - comparing the start position of the event to be executed. Since we do not have the start position available, we compute it by subtracting event length from end position (which is available). - if there are no events on the event queue at the slave sql starting time, that meet until condition, we stop immediately, as in this case we do not want to wait for next event.
2008-02-22 16:07:07 +01:00
/*
First check until condition - probably there is nothing to execute. We
do not want to wait for next event in this case.
*/
pthread_mutex_lock(&rli->data_lock);
BUG#13861 - START SLAVE UNTIL may stop 1 evnt too late if log-slave-updates and circul repl After merge fixes.
2008-02-27 18:46:06 +01:00
if (rli->until_condition != Relay_log_info::UNTIL_NONE &&
BUG#13861 - START SLAVE UNTIL may stop 1 evnt too late if log-slave-updates and circul repl Slave SQL thread may execute one extra event when there are events skipped by slave I/O thread (e.g. originated by the same server). Whereas it was requested not to do so by the UNTIL condition. This happens because we compare with the end position of previously executed event. This is fine when there are no skipped by slave I/O thread events, as end position of previous event equals to start position of to be executed event. Otherwise this position equals to start position of skipped event. This is fixed by: - reading the event to be executed before checking if the until condition is satisfied. - comparing the start position of the event to be executed. Since we do not have the start position available, we compute it by subtracting event length from end position (which is available). - if there are no events on the event queue at the slave sql starting time, that meet until condition, we stop immediately, as in this case we do not want to wait for next event.
2008-02-22 16:07:07 +01:00
rli->is_until_satisfied(rli->group_master_log_pos))
{
char buf[22];
sql_print_information("Slave SQL thread stopped because it reached its"
" UNTIL position %s", llstr(rli->until_pos(), buf));
pthread_mutex_unlock(&rli->data_lock);
goto err;
}
pthread_mutex_unlock(&rli->data_lock);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
/* Read queries from the IO/THREAD until this thread is killed */
fixed replication assert failure ( found by Jeremy Zawodny)
2002-03-08 23:02:11 +01:00
while (!sql_slave_killed(thd,rli))
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Reading event from the relay log");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(rli->sql_thd == thd);
replication updates. This changeset seems to be working fine on test systems. If no problems are discovered in the next week, this will make the replication code ready for 4.0.2 release.
2002-03-30 20:36:05 +01:00
THD_CHECK_SENTRY(thd);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (exec_relay_log_event(thd,rli))
{
BUG#23171 (Illegal slave restart position): Third patch of the bug fix where the code for skipping events and for executing events is factored out into three functions: - shall_skip() to decide if the event shall be skipped and the reason for it; - do_apply_event(), where the event is applied to the database; and - do_update_pos(), which updates the actual relay log position and group positions.
2007-01-17 15:06:37 +01:00
DBUG_PRINT("info", ("exec_relay_log_event() failed"));
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
// do not scare the user if SQL thread was simply killed or stopped
fixed replication assert failure ( found by Jeremy Zawodny)
2002-03-08 23:02:11 +01:00
if (!sql_slave_killed(thd,rli))
wl2325 wl2324
2006-01-12 19:51:02 +01:00
{
/*
BUG#23171 (Illegal slave restart position): Third patch of the bug fix where the code for skipping events and for executing events is factored out into three functions: - shall_skip() to decide if the event shall be skipped and the reason for it; - do_apply_event(), where the event is applied to the database; and - do_update_pos(), which updates the actual relay log position and group positions.
2007-01-17 15:06:37 +01:00
retrieve as much info as possible from the thd and, error
codes and warnings and print this to the error log as to
allow the user to locate the error
wl2325 wl2324
2006-01-12 19:51:02 +01:00
*/
BUG#24954 (Last_errno and Last_error not set after master_retry_count has been reached): Post-merge patch to handle all the changes to the tree since the tree was cloned.
2007-06-11 22:15:39 +02:00
uint32 const last_errno= rli->last_error().number;
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
Bug#12713 "Error in a stored function called from a SELECT doesn't cause ROLLBACK of statement", part 1. Review fixes. Do not send OK/EOF packets to the client until we reached the end of the current statement. This is a consolidation, to keep the functionality that is shared by all SQL statements in one place in the server. Currently this functionality includes: - close_thread_tables() - log_slow_statement(). After this patch and the subsequent patch for Bug#12713, it shall also include: - ha_autocommit_or_rollback() - net_end_statement() - query_cache_end_of_result(). In future it may also include: - mysql_reset_thd_for_next_command().
2007-12-12 16:21:01 +01:00
if (thd->is_error())
wl2325 wl2324
2006-01-12 19:51:02 +01:00
{
Bug#12713 "Error in a stored function called from a SELECT doesn't cause ROLLBACK of statement", part 1. Review fixes. Do not send OK/EOF packets to the client until we reached the end of the current statement. This is a consolidation, to keep the functionality that is shared by all SQL statements in one place in the server. Currently this functionality includes: - close_thread_tables() - log_slow_statement(). After this patch and the subsequent patch for Bug#12713, it shall also include: - ha_autocommit_or_rollback() - net_end_statement() - query_cache_end_of_result(). In future it may also include: - mysql_reset_thd_for_next_command().
2007-12-12 16:21:01 +01:00
char const *const errmsg= thd->main_da.message();
DBUG_PRINT("info",
("thd->main_da.sql_errno()=%d; rli->last_error.number=%d",
thd->main_da.sql_errno(), last_errno));
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
if (last_errno == 0)
wl2325 wl2324
2006-01-12 19:51:02 +01:00
{
BUG#38197 Errors in @@init_slave not visible in 'show slave status' Some errors that cause the slave SQL thread to stop are not shown in the Slave_SQL_Error column of "SHOW SLAVE STATUS". Instead, the error is only in the server's error log. That makes it difficult to analyze the error for the user. One example of an error that stops the slave but is not shown by "SHOW SLAVE STATUS" is when @@global.init_slave is set incorrectly (e.g., it contains something that is not valid SQL). Three failures were not correctly reported: 1 - Failures during slave thread initialization 2 - Failures while initializing the relay log position right after starting the slave thread. 3 - Failures while processing queries passed through the init_slave option. This patch fixes the issues by reporting the errors through relay-info->report.
2009-02-11 12:56:25 +01:00
/*
This function is reporting an error which was not reported
while executing exec_relay_log_event().
*/
Bug #43414 Parenthesis (and other) warnings compiling MySQL with gcc 4.3.2 Cleaning up warnings not present in 5.0.
2009-09-23 15:21:29 +02:00
rli->report(ERROR_LEVEL, thd->main_da.sql_errno(), "%s", errmsg);
wl2325 wl2324
2006-01-12 19:51:02 +01:00
}
Bug#12713 "Error in a stored function called from a SELECT doesn't cause ROLLBACK of statement", part 1. Review fixes. Do not send OK/EOF packets to the client until we reached the end of the current statement. This is a consolidation, to keep the functionality that is shared by all SQL statements in one place in the server. Currently this functionality includes: - close_thread_tables() - log_slow_statement(). After this patch and the subsequent patch for Bug#12713, it shall also include: - ha_autocommit_or_rollback() - net_end_statement() - query_cache_end_of_result(). In future it may also include: - mysql_reset_thd_for_next_command().
2007-12-12 16:21:01 +01:00
else if (last_errno != thd->main_da.sql_errno())
wl2325 wl2324
2006-01-12 19:51:02 +01:00
{
BUG#38197 Errors in @@init_slave not visible in 'show slave status' Some errors that cause the slave SQL thread to stop are not shown in the Slave_SQL_Error column of "SHOW SLAVE STATUS". Instead, the error is only in the server's error log. That makes it difficult to analyze the error for the user. One example of an error that stops the slave but is not shown by "SHOW SLAVE STATUS" is when @@global.init_slave is set incorrectly (e.g., it contains something that is not valid SQL). Three failures were not correctly reported: 1 - Failures during slave thread initialization 2 - Failures while initializing the relay log position right after starting the slave thread. 3 - Failures while processing queries passed through the init_slave option. This patch fixes the issues by reporting the errors through relay-info->report.
2009-02-11 12:56:25 +01:00
/*
* An error was reported while executing exec_relay_log_event()
* however the error code differs from what is in the thread.
* This function prints out more information to help finding
* what caused the problem.
*/
wl2325 wl2324
2006-01-12 19:51:02 +01:00
sql_print_error("Slave (additional info): %s Error_code: %d",
Bug#12713 "Error in a stored function called from a SELECT doesn't cause ROLLBACK of statement", part 1. Review fixes. Do not send OK/EOF packets to the client until we reached the end of the current statement. This is a consolidation, to keep the functionality that is shared by all SQL statements in one place in the server. Currently this functionality includes: - close_thread_tables() - log_slow_statement(). After this patch and the subsequent patch for Bug#12713, it shall also include: - ha_autocommit_or_rollback() - net_end_statement() - query_cache_end_of_result(). In future it may also include: - mysql_reset_thd_for_next_command().
2007-12-12 16:21:01 +01:00
errmsg, thd->main_da.sql_errno());
wl2325 wl2324
2006-01-12 19:51:02 +01:00
}
}
/* Print any warnings issued */
List_iterator_fast<MYSQL_ERROR> it(thd->warn_list);
MYSQL_ERROR *err;
WL#3629 - Replication of Invocation and Invoked Features This changeset adds replication of events and user-defined functions. There are several bug reports involved in this change: BUG#16421, BUG#17857, BUG#20384: This patch modifies the mysql.events table to permit the addition of another enum value for the status column. The column now has values of ('DISABLED','SLAVESIDE_DISABLED','ENABLED'). A status of SLAVESIDE_DISABLED is set on the slave during replication of events. This enables users to determine which events werereplicated from the master and to later enable them if they promote the slave to a master. The CREATE, ALTER, and DROP statements are binlogged. A new test was added for replication of events (rpl_events). BUG#17671: This patch modifies the code to permit logging of user-defined functions. Note: this is the CREATE FUNCTION ... SONAME variety. A more friendly error message to be displayed should a replicated user-defined function not be found in the loadable library or if the library is missing from the slave.The CREATE andDROP statements are binlogged. A new test was added for replication of user-defined functions (rpl_udf). The patch also adds a new column to the mysql.event table named 'originator' that is used to store the server_id of the server that the event originated on. This enables users to promote a slave to a master and later return the promoted slave to a slave and disable the replicated events.
2007-03-16 14:56:57 +01:00
/*
Added controlled slave thread cancel for replication
of user-defined variables.
*/
bool udf_error = false;
wl2325 wl2324
2006-01-12 19:51:02 +01:00
while ((err= it++))
WL#3629 - Replication of Invocation and Invoked Features This changeset adds replication of events and user-defined functions. There are several bug reports involved in this change: BUG#16421, BUG#17857, BUG#20384: This patch modifies the mysql.events table to permit the addition of another enum value for the status column. The column now has values of ('DISABLED','SLAVESIDE_DISABLED','ENABLED'). A status of SLAVESIDE_DISABLED is set on the slave during replication of events. This enables users to determine which events werereplicated from the master and to later enable them if they promote the slave to a master. The CREATE, ALTER, and DROP statements are binlogged. A new test was added for replication of events (rpl_events). BUG#17671: This patch modifies the code to permit logging of user-defined functions. Note: this is the CREATE FUNCTION ... SONAME variety. A more friendly error message to be displayed should a replicated user-defined function not be found in the loadable library or if the library is missing from the slave.The CREATE andDROP statements are binlogged. A new test was added for replication of user-defined functions (rpl_udf). The patch also adds a new column to the mysql.event table named 'originator' that is used to store the server_id of the server that the event originated on. This enables users to promote a slave to a master and later return the promoted slave to a slave and disable the replicated events.
2007-03-16 14:56:57 +01:00
{
if (err->code == ER_CANT_OPEN_LIBRARY)
udf_error = true;
wl2325 wl2324
2006-01-12 19:51:02 +01:00
sql_print_warning("Slave: %s Error_code: %d",err->msg, err->code);
WL#3629 - Replication of Invocation and Invoked Features This changeset adds replication of events and user-defined functions. There are several bug reports involved in this change: BUG#16421, BUG#17857, BUG#20384: This patch modifies the mysql.events table to permit the addition of another enum value for the status column. The column now has values of ('DISABLED','SLAVESIDE_DISABLED','ENABLED'). A status of SLAVESIDE_DISABLED is set on the slave during replication of events. This enables users to determine which events werereplicated from the master and to later enable them if they promote the slave to a master. The CREATE, ALTER, and DROP statements are binlogged. A new test was added for replication of events (rpl_events). BUG#17671: This patch modifies the code to permit logging of user-defined functions. Note: this is the CREATE FUNCTION ... SONAME variety. A more friendly error message to be displayed should a replicated user-defined function not be found in the loadable library or if the library is missing from the slave.The CREATE andDROP statements are binlogged. A new test was added for replication of user-defined functions (rpl_udf). The patch also adds a new column to the mysql.event table named 'originator' that is used to store the server_id of the server that the event originated on. This enables users to promote a slave to a master and later return the promoted slave to a slave and disable the replicated events.
2007-03-16 14:56:57 +01:00
}
if (udf_error)
sql_print_error("Error loading user-defined library, slave SQL "
"thread aborted. Install the missing library, and restart the "
"slave SQL thread with \"SLAVE START\". We stopped at log '%s' "
"position %s", RPL_LOG_NAME, llstr(rli->group_master_log_pos,
llbuff));
else
sql_print_error("\
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
Error running query, slave SQL thread aborted. Fix the problem, and restart \
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
the slave SQL thread with \"SLAVE START\". We stopped at log \
few s/sql_print_error/sql_print_information/
2004-09-08 10:45:50 +02:00
'%s' position %s", RPL_LOG_NAME, llstr(rli->group_master_log_pos, llbuff));
wl2325 wl2324
2006-01-12 19:51:02 +01:00
}
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
goto err;
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
}
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
/* Thread stopped. Print the current replication position to the log */
merge with 4.1
2004-10-29 18:26:52 +02:00
sql_print_information("Slave SQL thread exiting, replication stopped in log "
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
"'%s' at position %s",
RPL_LOG_NAME, llstr(rli->group_master_log_pos,llbuff));
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
err:
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
/*
Some events set some playgrounds, which won't be cleared because thread
stops. Stopping of this thread may not be known to these events ("stop"
request is detected only by the present function, not by events), so we
must "proactively" clear playgrounds:
*/
rli->cleanup_context(thd, 1);
This is the final commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) It contains a few bugfixes (which I made when running the testsuite). I carefully updated the results of the testsuite (i.e. I checked for every one, if the difference between .reject and .result could be explained). Apparently mysql-test-run --manager is broken in 4.1 and 5.0 currently, so I could neither run the few tests which require --manager, nor check that they pass nor modify their .result. But for builds, we don't run with --manager. Apart from --manager, the full testsuite passes, with Valgrind too (no errors). I'm going to push in the next minutes. Remains: update the manual. Note: by chance I saw that (in 4.1, in 5.0) rpl_get_lock fails when run alone; this is normal at it makes assumptions on thread ids. I will fix this one day in 4.1.
2003-12-19 22:40:23 +01:00
/*
Some extra safety, which should not been needed (normally, event deletion
should already have done these assignments (each event which sets these
variables is supposed to set them to 0 before terminating)).
*/
Bug #38816: kill + flush tables with read lock + stored procedures causes crashes! The problem of that bugreport was mostly fixed by the patch for bug 38691. However, attached test case focused on another crash or valgrind warning problem: SHOW PROCESSLIST query accesses freed memory of SP instruction that run in a parallel connection. Changes of thd->query/thd->query_length in dangerous places have been guarded with the per-thread LOCK_thd_data mutex (the THD::LOCK_delete mutex has been renamed to THD::LOCK_thd_data).
2009-07-24 17:58:58 +02:00
thd->catalog= 0;
thd->set_query(NULL, 0);
A fix and a test case for Bug#19022 "Memory bug when switching db during trigger execution" Bug#17199 "Problem when view calls function from another database." Bug#18444 "Fully qualified stored function names don't work correctly in SELECT statements" Documentation note: this patch introduces a change in behaviour of prepared statements. This patch adds a few new invariants with regard to how THD::db should be used. These invariants should be preserved in future: - one should never refer to THD::db by pointer and always make a deep copy (strmake, strdup) - one should never compare two databases by pointer, but use strncmp or my_strncasecmp - TABLE_LIST object table->db should be always initialized in the parser or by creator of the object. For prepared statements it means that if the current database is changed after a statement is prepared, the database that was current at prepare remains active. This also means that you can not prepare a statement that implicitly refers to the current database if the latter is not set. This is not documented, and therefore needs documentation. This is NOT a change in behavior for almost all SQL statements except: - ALTER TABLE t1 RENAME t2 - OPTIMIZE TABLE t1 - ANALYZE TABLE t1 - TRUNCATE TABLE t1 -- until this patch t1 or t2 could be evaluated at the first execution of prepared statement. CURRENT_DATABASE() still works OK and is evaluated at every execution of prepared statement. Note, that in stored routines this is not an issue as the default database is the database of the stored procedure and "use" statement is prohibited in stored routines. This patch makes obsolete the use of check_db_used (it was never used in the old code too) and all other places that check for table->db and assign it from THD::db if it's NULL, except the parser. How this patch was created: THD::{db,db_length} were replaced with a LEX_STRING, THD::db. All the places that refer to THD::{db,db_length} were manually checked and: - if the place uses thd->db by pointer, it was fixed to make a deep copy - if a place compared two db pointers, it was fixed to compare them by value (via strcmp/my_strcasecmp, whatever was approproate) Then this intermediate patch was used to write a smaller patch that does the same thing but without a rename. TODO in 5.1: - remove check_db_used - deploy THD::set_db in mysql_change_db See also comments to individual files.
2006-06-26 22:47:52 +02:00
thd->reset_db(NULL, 0);
Prevent bugs by making DBUG_* expressions syntactically equivalent to a single statement. --- Bug#24795: SHOW PROFILE Profiling is only partially functional on some architectures. Where there is no getrusage() system call, presently Null values are returned where it would be required. Notably, Windows needs some love applied to make it as useful. Syntax this adds: SHOW PROFILES SHOW PROFILE [types] [FOR QUERY n] [OFFSET n] [LIMIT n] where "n" is an integer and "types" is zero or many (comma-separated) of "CPU" "MEMORY" (not presently supported) "BLOCK IO" "CONTEXT SWITCHES" "PAGE FAULTS" "IPC" "SWAPS" "SOURCE" "ALL" It also adds a session variable (boolean) "profiling", set to "no" by default, and (integer) profiling_history_size, set to 15 by default. This patch abstracts setting THDs' "proc_info" behind a macro that can be used as a hook into the profiling code when profiling support is compiled in. All future code in this line should use that mechanism for setting thd->proc_info. --- Tests are now set to omit the statistics. --- Adds an Information_schema table, "profiling" for access to "show profile" data. --- Merge zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community-3--bug24795 into zippy.cornsilk.net:/home/cmiller/work/mysql/mysql-5.0-community --- Fix merge problems. --- Fixed one bug in the query_source being NULL. Updated test results. --- Include more thorough profiling tests. Improve support for prepared statements. Use session-specific query IDs, starting at zero. --- Selecting from I_S.profiling is no longer quashed in profiling, as requested by Giuseppe. Limit the size of captured query text. No longer log queries that are zero length.
2007-02-22 16:03:08 +01:00
thd_proc_info(thd, "Waiting for slave mutex on exit");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_mutex_lock(&rli->run_lock);
- Fix for bug 651: now a dying SQL slave threads wakes up any waiting MASTER_POS_WAIT(). Could not add a testcase for this: if the test goes into a MASTER_POS_WAIT, it waits until this terminates (even doing "connection other_con" to launch "stop slave" is blocked). - In MASTER_POS_WAIT() don't test if the I/O slave is running, but if the SQL thread is running. - Some DBUG info for this bugfix.
2003-06-14 16:40:00 +02:00
/* We need data_lock, at least to wake up any waiting master_pos_wait() */
pthread_mutex_lock(&rli->data_lock);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(rli->slave_running == 1); // tracking buffer overrun
- Fix for bug 651: now a dying SQL slave threads wakes up any waiting MASTER_POS_WAIT(). Could not add a testcase for this: if the test goes into a MASTER_POS_WAIT, it waits until this terminates (even doing "connection other_con" to launch "stop slave" is blocked). - In MASTER_POS_WAIT() don't test if the I/O slave is running, but if the SQL thread is running. - Some DBUG info for this bugfix.
2003-06-14 16:40:00 +02:00
/* When master_pos_wait() wakes up it will check this and terminate */
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
rli->slave_running= 0;
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
/* Forget the relay log's format */
delete rli->relay_log.description_event_for_exec;
rli->relay_log.description_event_for_exec= 0;
- Fix for bug 651: now a dying SQL slave threads wakes up any waiting MASTER_POS_WAIT(). Could not add a testcase for this: if the test goes into a MASTER_POS_WAIT, it waits until this terminates (even doing "connection other_con" to launch "stop slave" is blocked). - In MASTER_POS_WAIT() don't test if the I/O slave is running, but if the SQL thread is running. - Some DBUG info for this bugfix.
2003-06-14 16:40:00 +02:00
/* Wake up master_pos_wait() */
pthread_mutex_unlock(&rli->data_lock);
DBUG_PRINT("info",("Signaling possibly waiting master_pos_wait() functions"));
pthread_cond_broadcast(&rli->data_cond);
Simplified a test. thd->enter_cond() and exit_cond(), so that the I/O thread accepts to stop when it's waiting for relay log space. Reset ignore_log_space_limit to 0 when the SQL thread terminates.
2003-06-15 12:01:51 +02:00
rli->ignore_log_space_limit= 0; /* don't need any lock */
WL#1062 "log charset info into all Query_log_event": we store 7 bytes (1 + 2*3) in every Query_log_event. In the future if users want binlog optimized for small size and less safe, we could add --binlog-no-charset (and binlog-no-sql-mode etc): charset info is something by design optional (even if for now we don't offer possibility to disable it): it's not a binlog format change. We try to reduce the number of get_charset() calls in the slave SQL thread to a minimum by caching the charset read from the previous event (which will often be equal to the one of the current event). We don't use SET ONE_SHOT for charset-aware repl (we still do for timezones, will be fixed later). No more errors if one changes the global value of charset vars on master or slave (as we log charset info in all Query_log_event). Not fixing Load_log_event as it will be rewritten soon by Dmitri. Testing how mysqlbinlog behaves in rpl_charset.test. mysqlbinlog needs to know where charset file is (to be able to convert a charset number found in binlog (e.g. in User_var_log_event) to a charset name); mysql-test-run needs to pass the correct value for this option to mysqlbinlog. Many result udpates (adding charset info into every event shifts log_pos in SHOW BINLOG EVENTS). Roughly the same job is to be done for timezones :)
2005-02-03 16:22:16 +01:00
/* we die so won't remember charset - re-update them on next thread start */
rli->cached_charset_invalidate();
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
rli->save_temporary_tables = thd->temporary_tables;
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
/*
TODO: see if we can do this conditionally in next_event() instead
to avoid unneeded position re-init
*/
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
thd->temporary_tables = 0; // remove tempation from destructor to close them
DBUG_ASSERT(thd->net.buff != 0);
net_end(&thd->net); // destructor will not free it, because we are weird
DBUG_ASSERT(rli->sql_thd == thd);
replication updates. This changeset seems to be working fine on test systems. If no problems are discovered in the next week, this will make the replication code ready for 4.0.2 release.
2002-03-30 20:36:05 +01:00
THD_CHECK_SENTRY(thd);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
rli->sql_thd= 0;
BUG#41725: slave crashes when inserting into temporary table after stop/start slave When stopping and restarting the slave while it is replicating temporary tables, the server would crash or raise an assertion failure. This was due to the fact that although temporary tables are saved between slave threads restart, the reference to the thread in use (table->in_use) was not being properly updated when the restart happened (it would still reference the old/invalid thread instead of the new one). This patch addresses this issue by resetting the reference to the new slave thread on slave thread restart.
2009-05-23 01:15:21 +02:00
set_thd_in_use_temporary_tables(rli); // (re)set sql_thd in use for saved temp tables
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_mutex_lock(&LOCK_thread_count);
replication updates. This changeset seems to be working fine on test systems. If no problems are discovered in the next week, this will make the replication code ready for 4.0.2 release.
2002-03-30 20:36:05 +01:00
THD_CHECK_SENTRY(thd);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
delete thd;
pthread_mutex_unlock(&LOCK_thread_count);
Merge quant.(none):/ext/mysql/bk/mysql-5.0-bug25306 into quant.(none):/ext/mysql/bkroot/mysql-5.1-new-rpl
2007-03-01 10:03:42 +01:00
/*
Note: the order of the broadcast and unlock calls below (first broadcast, then unlock)
is important. Otherwise a killer_thread can execute between the calls and
delete the mi structure leading to a crash! (see BUG#25306 for details)
*/
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_cond_broadcast(&rli->stop_cond);
Bug #38694 Race condition in replication thread shutdown The issue of the current bug is unguarded access to mi->slave_running by the shutdown thread calling end_slave() that is bug#29968 (alas happened not to be cross-linked with the current bug) Fixed: with removing the unguarded read of the running status and perform reading it in terminate_slave_thread() at time run_lock is taken (mostly bug#29968 backporting, still with some improvements over that patch - see the error reporting from terminate_slave_thread()). Issue of bug#38716 is fixed here for 5.0 branch as well. Note: There has been a separate artifact identified - a race condition between init_slave() and end_slave() - reported as Bug#44467.
2009-04-28 13:46:07 +02:00
DBUG_EXECUTE_IF("simulate_slave_delay_at_terminate_bug38694", sleep(5););
BUG#25306 (Race conditions during replication slave shutdown (valgrind stacks)): The possibility of the race is removed by changing sequence of calls pthread_mutex_unlock(&mi->run_lock); pthread_cond_broadcast(&mi->stop_cond); into pthread_cond_broadcast(&mi->stop_cond); pthread_mutex_unlock(&mi->run_lock); at the end of I/O thread (similar change at the end of SQL thread). This ensures that no thread waiting on the condition executes between the broadcast and the unlock and thus can't delete the mi structure which caused the bug.
2007-02-03 20:14:16 +01:00
pthread_mutex_unlock(&rli->run_lock); // tell the world we are done
A patch for Bug#47474 (mysqld hits Dbug_violation_helper assert when compiled with Sun Studio compiler). The thing is that Sun Studio compiler calls destructor of stack objects when pthread_exit() is called. That triggered an assertion in DBUG_ENTER()/DBUG_RETURN() validation logic (if DBUG_ENTER() is used in the beginning of function, all returns should be replaced by DBUG_RETURN/DBUG_VOID_RETURN macros). A fix is to explicitly use DBUG_LEAVE macro.
2009-09-23 15:10:23 +02:00
DBUG_LEAVE; // Must match DBUG_ENTER()
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
my_thread_end();
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
pthread_exit(0);
A patch for Bug#47474 (mysqld hits Dbug_violation_helper assert when compiled with Sun Studio compiler). The thing is that Sun Studio compiler calls destructor of stack objects when pthread_exit() is called. That triggered an assertion in DBUG_ENTER()/DBUG_RETURN() validation logic (if DBUG_ENTER() is used in the beginning of function, all returns should be replaced by DBUG_RETURN/DBUG_VOID_RETURN macros). A fix is to explicitly use DBUG_LEAVE macro.
2009-09-23 15:10:23 +02:00
return 0; // Avoid compiler warnings
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
keep bk happy
2001-01-22 03:46:32 +01:00
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
/*
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
process_io_create_file()
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
*/
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int process_io_create_file(Master_info* mi, Create_file_log_event* cev)
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
{
int error = 1;
ulong num_bytes;
bool cev_not_written;
Fixes and code cleanups after merge with 4.0.3 Warning handling and initial prepared statement handling (last not complete yet) Changed a lot of functions that returned 0/1 to my_bool type. GRANT handling now uses read/write locks instead of mutex Change basic net functions to use THD instead of NET (needed for 4.1 protocol) Use my_sprintf instead of sprintf() + strlen() Added alloc_query() to be able to chare query initialization code with prepared statements. Cleanup handling of SHOW COUNT(*) WARNINGS and SELECT LAST_INSERT_ID() Note that the following test fails (will be fixed ASAP): sub_select, union, rpl_rotate_logs and rpl_mystery22
2002-10-02 12:33:08 +02:00
THD *thd = mi->io_thd;
NET *net = &mi->mysql->net;
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_ENTER("process_io_create_file");
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
if (unlikely(!cev->is_valid()))
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(1);
WL#2472: Refactoring of replication and binlog filtering. Needed to be able to have multiple masters connected to same slave server with different filters.
2005-03-08 21:12:35 +01:00
if (!rpl_filter->db_ok(cev->db))
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
{
skip_load_data_infile(net);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(0);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
}
DBUG_ASSERT(cev->inited_from_old);
thd->file_id = cev->file_id = mi->file_id++;
updated test suite to be able to run 3.23 master and 4.0 slave got LOAD DATA INFILE replication from 3.23 master to 4.0 slave to work for some reason query_cache now fails - will push anyway and document the failure if I can repeat it.
2002-02-09 21:58:53 +01:00
thd->server_id = cev->server_id;
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
cev_not_written = 1;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
if (unlikely(net_request_file(net,cev->fname)))
{
sql_print_error("Slave I/O: failed requesting download of '%s'",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
cev->fname);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
goto err;
}
Fixes for Netware Call pthread_mutex_destroy() on not used mutex. Changed comments in .h and .c files from // -> /* */ Added detection of mutex on which one didn't call pthread_mutex_destroy() Fixed bug in create_tmp_field() which causes a memory overrun in queries that uses "ORDER BY constant_expression" Added optimisation for ORDER BY NULL
2003-01-28 07:38:28 +01:00
/*
This dummy block is so we could instantiate Append_block_log_event
once and then modify it slightly instead of doing it multiple times
in the loop
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
*/
{
Yesterday I removed process_io_create_file; I shouldn't have. Let's say the lack of comments did not help me ;) Copying it back again and adding comments; now 3.23->4.0 replication of LOAD DATA INFILE works again.
2003-08-21 10:24:37 +02:00
Append_block_log_event aev(thd,0,0,0,0);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
for (;;)
{
if (unlikely((num_bytes=my_net_read(net)) == packet_error))
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
sql_print_error("Network read error downloading '%s' from master",
cev->fname);
goto err;
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
}
if (unlikely(!num_bytes)) /* eof */
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
/* 3.23 master wants it */
net_write_command(net, 0, (uchar*) "", 0, (uchar*) "", 0);
Merge with 4.0.18
2004-02-11 00:06:46 +01:00
/*
If we wrote Create_file_log_event, then we need to write
Execute_load_log_event. If we did not write Create_file_log_event,
then this is an empty file and we can just do as if the LOAD DATA
INFILE had not existed, i.e. write nothing.
*/
if (unlikely(cev_not_written))
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
break;
Execute_load_log_event xev(thd,0,0);
xev.log_pos = cev->log_pos;
if (unlikely(mi->rli.relay_log.append(&xev)))
{
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
mi->report(ERROR_LEVEL, ER_SLAVE_RELAY_LOG_WRITE_FAILURE,
ER(ER_SLAVE_RELAY_LOG_WRITE_FAILURE),
"error writing Exec_load event to relay log");
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
}
mi->rli.relay_log.harvest_bytes_written(&mi->rli.log_space_total);
break;
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
}
if (unlikely(cev_not_written))
{
Post-merge changes.
2008-01-30 16:03:00 +01:00
cev->block = net->read_pos;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
cev->block_len = num_bytes;
if (unlikely(mi->rli.relay_log.append(cev)))
{
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
mi->report(ERROR_LEVEL, ER_SLAVE_RELAY_LOG_WRITE_FAILURE,
ER(ER_SLAVE_RELAY_LOG_WRITE_FAILURE),
"error writing Create_file event to relay log");
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
}
cev_not_written=0;
mi->rli.relay_log.harvest_bytes_written(&mi->rli.log_space_total);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
}
else
{
Post-merge changes.
2008-01-30 16:03:00 +01:00
aev.block = net->read_pos;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
aev.block_len = num_bytes;
aev.log_pos = cev->log_pos;
if (unlikely(mi->rli.relay_log.append(&aev)))
{
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
mi->report(ERROR_LEVEL, ER_SLAVE_RELAY_LOG_WRITE_FAILURE,
ER(ER_SLAVE_RELAY_LOG_WRITE_FAILURE),
"error writing Append_block event to relay log");
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
}
mi->rli.relay_log.harvest_bytes_written(&mi->rli.log_space_total) ;
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
}
}
}
error=0;
err:
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(error);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
}
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
Start using a new binary log on the master
SYNOPSIS
process_io_rotate()
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mi master_info for the slave
rev The rotate log event read from the binary log
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
DESCRIPTION
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
Updates the master info with the place in the next binary
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
log where we should start reading.
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
Rotate the relay log to avoid mixed-format relay logs.
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
NOTES
We assume we already locked mi->data_lock
RETURN VALUES
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
0 ok
1 Log event is illegal
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int process_io_rotate(Master_info *mi, Rotate_log_event *rev)
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
{
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
DBUG_ENTER("process_io_rotate");
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
safe_mutex_assert_owner(&mi->data_lock);
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
if (unlikely(!rev->is_valid()))
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
DBUG_RETURN(1);
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
/* Safe copy as 'rev' has been "sanitized" in Rotate_log_event's ctor */
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
memcpy(mi->master_log_name, rev->new_log_ident, rev->ident_len+1);
mi->master_log_pos= rev->pos;
Remove compiler warnings (Mostly in DBUG_PRINT() and unused arguments) Fixed bug in query cache when used with traceing (--with-debug) Fixed memory leak in mysqldump Removed warnings from mysqltest scripts (replaced -- with #)
2006-11-20 21:42:06 +01:00
DBUG_PRINT("info", ("master_log_pos: '%s' %lu",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mi->master_log_name, (ulong) mi->master_log_pos));
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
#ifndef DBUG_OFF
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
/*
If we do not do this, we will be getting the first
rotate event forever, so we need to not disconnect after one.
*/
if (disconnect_slave_event_count)
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
mi->events_till_disconnect++;
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
#endif
auto-ROLLBACK if binlog was not closed properly auto-commit on Xid_log_event
2005-02-09 20:04:28 +01:00
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
/*
If description_event_for_queue is format <4, there is conversion in the
relay log to the slave's format (4). And Rotate can mean upgrade or
nothing. If upgrade, it's to 5.0 or newer, so we will get a Format_desc, so
no need to reset description_event_for_queue now. And if it's nothing (same
master version as before), no need (still using the slave's format).
*/
if (mi->rli.relay_log.description_event_for_queue->binlog_version >= 4)
{
delete mi->rli.relay_log.description_event_for_queue;
/* start from format 3 (MySQL 4.0) again */
mi->rli.relay_log.description_event_for_queue= new
Format_description_log_event(3);
}
Replication: various small fixes specific to the new binlog format of 5.0 (including one which may explain autobuild's failure of yesterday)
2004-07-26 19:42:59 +02:00
/*
Rotate the relay log makes binlog format detection easier (at next slave
start or mysqlbinlog)
*/
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
rotate_relay_log(mi); /* will take the right mutexes */
Added a lot of DBUG_xxx statements to be able to find replication bug. Fixed critical bug on 64 bit systems. Cleanups
2002-06-08 20:02:01 +02:00
DBUG_RETURN(0);
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
}
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
/*
auto-ROLLBACK if binlog was not closed properly auto-commit on Xid_log_event
2005-02-09 20:04:28 +01:00
Reads a 3.23 event and converts it to the slave's format. This code was
copied from MySQL 4.0.
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int queue_binlog_ver_1_event(Master_info *mi, const char *buf,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
ulong event_len)
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
{
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
const char *errmsg = 0;
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
ulong inc_pos;
bool ignore_event= 0;
char *tmp_buf = 0;
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
Relay_log_info *rli= &mi->rli;
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
DBUG_ENTER("queue_binlog_ver_1_event");
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
/*
If we get Load event, we need to pass a non-reusable buffer
to read_log_event, so we do a trick
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
*/
if (buf[EVENT_TYPE_OFFSET] == LOAD_EVENT)
{
if (unlikely(!(tmp_buf=(char*)my_malloc(event_len+1,MYF(MY_WME)))))
{
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
mi->report(ERROR_LEVEL, ER_SLAVE_FATAL_ERROR,
ER(ER_SLAVE_FATAL_ERROR), "Memory allocation failed");
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(1);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
}
memcpy(tmp_buf,buf,event_len);
Fix for BUG#3422 "In 3.23 -> 4.0 replication, slave segfault when replicating LOAD DATA INFILE": as we transform the 3.23 Load_log_event into a 4.0 Create_file_log_event which is one byte longer, we need to increment event_len. The bug was that we did not increment it, so later in code the end 0 was not seen so there was for example a segfault in strlen(fname) because fname was not 0-terminated. Other problems remain in 3.23->4.0 replication of LOAD DATA INFILE but they are less serious: Exec_master_log_pos and Relay_log_space are incorrect. I'll document them. They are not fixable without significant code changes (if you fix those problems in 4.0, you get assertion failures somewhere else etc), * which are already done in 5.0.0 *.
2004-04-08 22:12:22 +02:00
/*
Create_file constructor wants a 0 as last char of buffer, this 0 will
serve as the string-termination char for the file's name (which is at the
end of the buffer)
We must increment event_len, otherwise the event constructor will not see
this end 0, which leads to segfault.
*/
tmp_buf[event_len++]=0;
compatibility fix (bug#3606)
2004-04-30 10:42:12 +02:00
int4store(tmp_buf+EVENT_LEN_OFFSET, event_len);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
buf = (const char*)tmp_buf;
}
Yesterday I removed process_io_create_file; I shouldn't have. Let's say the lack of comments did not help me ;) Copying it back again and adding comments; now 3.23->4.0 replication of LOAD DATA INFILE works again.
2003-08-21 10:24:37 +02:00
/*
This will transform LOAD_EVENT into CREATE_FILE_EVENT, ask the master to
send the loaded file, and write it to the relay log in the form of
Append_block/Exec_load (the SQL thread needs the data, as that thread is not
connected to the master).
*/
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
Log_event *ev = Log_event::read_log_event(buf,event_len, &errmsg,
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
mi->rli.relay_log.description_event_for_queue);
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
if (unlikely(!ev))
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
{
sql_print_error("Read invalid event from master: '%s',\
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
master could be corrupt but a more likely cause of this is a bug",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
errmsg);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
my_free((char*) tmp_buf, MYF(MY_ALLOW_ZERO_PTR));
DBUG_RETURN(1);
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
}
BUG#23171 (Illegal slave restart position): Third patch of the bug fix where the code for skipping events and for executing events is factored out into three functions: - shall_skip() to decide if the event shall be skipped and the reason for it; - do_apply_event(), where the event is applied to the database; and - do_update_pos(), which updates the actual relay log position and group positions.
2007-01-17 15:06:37 +01:00
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
pthread_mutex_lock(&mi->data_lock);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
ev->log_pos= mi->master_log_pos; /* 3.23 events don't contain log_pos */
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
switch (ev->get_type_code()) {
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
case STOP_EVENT:
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
ignore_event= 1;
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
inc_pos= event_len;
break;
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
case ROTATE_EVENT:
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
if (unlikely(process_io_rotate(mi,(Rotate_log_event*)ev)))
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
{
delete ev;
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
pthread_mutex_unlock(&mi->data_lock);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(1);
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
}
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
inc_pos= 0;
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
break;
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
case CREATE_FILE_EVENT:
Yesterday I removed process_io_create_file; I shouldn't have. Let's say the lack of comments did not help me ;) Copying it back again and adding comments; now 3.23->4.0 replication of LOAD DATA INFILE works again.
2003-08-21 10:24:37 +02:00
/*
Yes it's possible to have CREATE_FILE_EVENT here, even if we're in
queue_old_event() which is for 3.23 events which don't comprise
CREATE_FILE_EVENT. This is because read_log_event() above has just
transformed LOAD_EVENT into CREATE_FILE_EVENT.
*/
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
{
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
/* We come here when and only when tmp_buf != 0 */
Remove compiler warnings and remove not used variables (Found during build process)
2005-02-25 15:53:22 +01:00
DBUG_ASSERT(tmp_buf != 0);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
inc_pos=event_len;
ev->log_pos+= inc_pos;
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
int error = process_io_create_file(mi,(Create_file_log_event*)ev);
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
delete ev;
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
mi->master_log_pos += inc_pos;
Remove compiler warnings (Mostly in DBUG_PRINT() and unused arguments) Fixed bug in query cache when used with traceing (--with-debug) Fixed memory leak in mysqldump Removed warnings from mysqltest scripts (replaced -- with #)
2006-11-20 21:42:06 +01:00
DBUG_PRINT("info", ("master_log_pos: %lu", (ulong) mi->master_log_pos));
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
pthread_mutex_unlock(&mi->data_lock);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
my_free((char*)tmp_buf, MYF(0));
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(error);
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
}
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
default:
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
inc_pos= event_len;
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
break;
}
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
if (likely(!ignore_event))
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
if (ev->log_pos)
/*
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
Don't do it for fake Rotate events (see comment in
Log_event::Log_event(const char* buf...) in log_event.cc).
*/
ev->log_pos+= event_len; /* make log_pos be the pos of the end of the event */
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
if (unlikely(rli->relay_log.append(ev)))
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
{
delete ev;
pthread_mutex_unlock(&mi->data_lock);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(1);
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
}
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
rli->relay_log.harvest_bytes_written(&rli->log_space_total);
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
}
delete ev;
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
mi->master_log_pos+= inc_pos;
Remove compiler warnings (Mostly in DBUG_PRINT() and unused arguments) Fixed bug in query cache when used with traceing (--with-debug) Fixed memory leak in mysqldump Removed warnings from mysqltest scripts (replaced -- with #)
2006-11-20 21:42:06 +01:00
DBUG_PRINT("info", ("master_log_pos: %lu", (ulong) mi->master_log_pos));
temporary commit - need to pull Monty's changes
2002-01-29 21:50:09 +01:00
pthread_mutex_unlock(&mi->data_lock);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(0);
more predicatable slave behaviour with wait_for_slave_stop in mysqltest fixed a couple of bugs with SEQ_READ_APPEND cache rpl000016 still has non-deterministic result, but I am going to commit and push since what I have is now better than what is in the main repository
2002-01-25 06:49:47 +01:00
}
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
/*
Reads a 4.0 event and converts it to the slave's format. This code was copied
from queue_binlog_ver_1_event(), with some affordable simplifications.
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int queue_binlog_ver_3_event(Master_info *mi, const char *buf,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
ulong event_len)
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
{
const char *errmsg = 0;
ulong inc_pos;
char *tmp_buf = 0;
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
Relay_log_info *rli= &mi->rli;
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
DBUG_ENTER("queue_binlog_ver_3_event");
/* read_log_event() will adjust log_pos to be end_log_pos */
Log_event *ev = Log_event::read_log_event(buf,event_len, &errmsg,
mi->rli.relay_log.description_event_for_queue);
if (unlikely(!ev))
{
sql_print_error("Read invalid event from master: '%s',\
master could be corrupt but a more likely cause of this is a bug",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
errmsg);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
my_free((char*) tmp_buf, MYF(MY_ALLOW_ZERO_PTR));
DBUG_RETURN(1);
}
pthread_mutex_lock(&mi->data_lock);
switch (ev->get_type_code()) {
case STOP_EVENT:
goto err;
case ROTATE_EVENT:
if (unlikely(process_io_rotate(mi,(Rotate_log_event*)ev)))
{
delete ev;
pthread_mutex_unlock(&mi->data_lock);
DBUG_RETURN(1);
}
inc_pos= 0;
break;
default:
inc_pos= event_len;
break;
}
if (unlikely(rli->relay_log.append(ev)))
{
delete ev;
pthread_mutex_unlock(&mi->data_lock);
DBUG_RETURN(1);
}
rli->relay_log.harvest_bytes_written(&rli->log_space_total);
delete ev;
mi->master_log_pos+= inc_pos;
err:
Remove compiler warnings (Mostly in DBUG_PRINT() and unused arguments) Fixed bug in query cache when used with traceing (--with-debug) Fixed memory leak in mysqldump Removed warnings from mysqltest scripts (replaced -- with #)
2006-11-20 21:42:06 +01:00
DBUG_PRINT("info", ("master_log_pos: %lu", (ulong) mi->master_log_pos));
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
pthread_mutex_unlock(&mi->data_lock);
DBUG_RETURN(0);
}
/*
queue_old_event()
Writes a 3.23 or 4.0 event to the relay log, after converting it to the 5.0
(exactly, slave's) format. To do the conversion, we create a 5.0 event from
the 3.23/4.0 bytes, then write this event to the relay log.
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
TODO:
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
Test this code before release - it has to be tested on a separate
setup with 3.23 master or 4.0 master
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int queue_old_event(Master_info *mi, const char *buf,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
ulong event_len)
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("queue_old_event");
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
switch (mi->rli.relay_log.description_event_for_queue->binlog_version)
{
case 1:
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(queue_binlog_ver_1_event(mi,buf,event_len));
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
case 3:
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(queue_binlog_ver_3_event(mi,buf,event_len));
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
default: /* unsupported format; eg version 2 */
DBUG_PRINT("info",("unsupported binlog format %d in queue_old_event()",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mi->rli.relay_log.description_event_for_queue->binlog_version));
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(1);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
}
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
/*
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
queue_event()
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
If the event is 3.23/4.0, passes it to queue_old_event() which will convert
it. Otherwise, writes a 5.0 (or newer) event to the relay log. Then there is
no format conversion, it's pure read/write of bytes.
So a 5.0.0 slave's relay log can contain events in the slave's format or in
any >=5.0.0 format.
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
*/
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
static int queue_event(Master_info* mi,const char* buf, ulong event_len)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
int error= 0;
ulong inc_pos;
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
Relay_log_info *rli= &mi->rli;
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
pthread_mutex_t *log_lock= rli->relay_log.get_log_lock();
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_ENTER("queue_event");
After merge fixes: - Remove compiler warnings - Fix merge errors - Indentation fixes
2006-12-15 05:21:15 +01:00
LINT_INIT(inc_pos);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
if (mi->rli.relay_log.description_event_for_queue->binlog_version<4 &&
buf[EVENT_TYPE_OFFSET] != FORMAT_DESCRIPTION_EVENT /* a way to escape */)
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(queue_old_event(mi,buf,event_len));
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
Fixed compiler and valgrind warnings Added missing DBUG_xxx_RETURN statements Fixed some usage of not initialized variables (as found by valgrind) Ensure that we don't remove locked tables used as name locks from open table cache until unlock_table_names() are called. This was fixed by having drop_locked_name() returning any table used as a name lock so that we can free it in unlock_table_names() This will allow Tomas to continue with his work to use namelocks to syncronize things. Note: valgrind still produces a lot of warnings about using not initialized code and shows memory loss errors when running the ndb tests
2006-03-29 13:27:36 +02:00
LINT_INIT(inc_pos);
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
pthread_mutex_lock(&mi->data_lock);
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
switch (buf[EVENT_TYPE_OFFSET]) {
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
case STOP_EVENT:
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
/*
We needn't write this event to the relay log. Indeed, it just indicates a
Fixes after merge
2003-10-08 11:01:58 +02:00
master server shutdown. The only thing this does is cleaning. But
cleaning is already done on a per-master-thread basis (as the master
server is shutting down cleanly, it has written all DROP TEMPORARY TABLE
First part of fix for BUG#7998 "Replication should be more clever about when to replicate RELEASE_LOCK()" (next part is updating test results)
2005-03-02 16:37:54 +01:00
prepared statements' deletion are TODO only when we binlog prep stmts).
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Fixes after merge
2003-10-08 11:01:58 +02:00
We don't even increment mi->master_log_pos, because we may be just after
a Rotate event. Btw, in a few milliseconds we are going to have a Start
event from the next binlog (unless the master is presently running
without --log-bin).
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
*/
goto err;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
case ROTATE_EVENT:
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Rotate_log_event rev(buf,event_len,mi->rli.relay_log.description_event_for_queue);
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
if (unlikely(process_io_rotate(mi,&rev)))
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
{
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
error= 1;
goto err;
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
}
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
/*
Now the I/O thread has just changed its mi->master_log_name, so
incrementing mi->master_log_pos is nonsense.
*/
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
inc_pos= 0;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
break;
}
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
case FORMAT_DESCRIPTION_EVENT:
{
/*
Create an event, and save it (when we rotate the relay log, we will have
to write this event again).
*/
/*
auto-ROLLBACK if binlog was not closed properly auto-commit on Xid_log_event
2005-02-09 20:04:28 +01:00
We are the only thread which reads/writes description_event_for_queue.
The relay_log struct does not move (though some members of it can
change), so we needn't any lock (no rli->data_lock, no log lock).
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
*/
This is the final commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) It contains a few bugfixes (which I made when running the testsuite). I carefully updated the results of the testsuite (i.e. I checked for every one, if the difference between .reject and .result could be explained). Apparently mysql-test-run --manager is broken in 4.1 and 5.0 currently, so I could neither run the few tests which require --manager, nor check that they pass nor modify their .result. But for builds, we don't run with --manager. Apart from --manager, the full testsuite passes, with Valgrind too (no errors). I'm going to push in the next minutes. Remains: update the manual. Note: by chance I saw that (in 4.1, in 5.0) rpl_get_lock fails when run alone; this is normal at it makes assumptions on thread ids. I will fix this one day in 4.1.
2003-12-19 22:40:23 +01:00
Format_description_log_event* tmp;
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
const char* errmsg;
This is the final commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) It contains a few bugfixes (which I made when running the testsuite). I carefully updated the results of the testsuite (i.e. I checked for every one, if the difference between .reject and .result could be explained). Apparently mysql-test-run --manager is broken in 4.1 and 5.0 currently, so I could neither run the few tests which require --manager, nor check that they pass nor modify their .result. But for builds, we don't run with --manager. Apart from --manager, the full testsuite passes, with Valgrind too (no errors). I'm going to push in the next minutes. Remains: update the manual. Note: by chance I saw that (in 4.1, in 5.0) rpl_get_lock fails when run alone; this is normal at it makes assumptions on thread ids. I will fix this one day in 4.1.
2003-12-19 22:40:23 +01:00
if (!(tmp= (Format_description_log_event*)
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
Log_event::read_log_event(buf, event_len, &errmsg,
This is the final commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) It contains a few bugfixes (which I made when running the testsuite). I carefully updated the results of the testsuite (i.e. I checked for every one, if the difference between .reject and .result could be explained). Apparently mysql-test-run --manager is broken in 4.1 and 5.0 currently, so I could neither run the few tests which require --manager, nor check that they pass nor modify their .result. But for builds, we don't run with --manager. Apart from --manager, the full testsuite passes, with Valgrind too (no errors). I'm going to push in the next minutes. Remains: update the manual. Note: by chance I saw that (in 4.1, in 5.0) rpl_get_lock fails when run alone; this is normal at it makes assumptions on thread ids. I will fix this one day in 4.1.
2003-12-19 22:40:23 +01:00
mi->rli.relay_log.description_event_for_queue)))
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
{
error= 2;
goto err;
}
This is the final commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) It contains a few bugfixes (which I made when running the testsuite). I carefully updated the results of the testsuite (i.e. I checked for every one, if the difference between .reject and .result could be explained). Apparently mysql-test-run --manager is broken in 4.1 and 5.0 currently, so I could neither run the few tests which require --manager, nor check that they pass nor modify their .result. But for builds, we don't run with --manager. Apart from --manager, the full testsuite passes, with Valgrind too (no errors). I'm going to push in the next minutes. Remains: update the manual. Note: by chance I saw that (in 4.1, in 5.0) rpl_get_lock fails when run alone; this is normal at it makes assumptions on thread ids. I will fix this one day in 4.1.
2003-12-19 22:40:23 +01:00
delete mi->rli.relay_log.description_event_for_queue;
mi->rli.relay_log.description_event_for_queue= tmp;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
/*
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
Though this does some conversion to the slave's format, this will
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
preserve the master's binlog format version, and number of event types.
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
*/
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
/*
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
If the event was not requested by the slave (the slave did not ask for
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
it), i.e. has end_log_pos=0, we do not increment mi->master_log_pos
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
*/
inc_pos= uint4korr(buf+LOG_POS_OFFSET) ? event_len : 0;
DBUG_PRINT("info",("binlog format is now %d",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mi->rli.relay_log.description_event_for_queue->binlog_version));
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
}
break;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
default:
Portability fixes. Improve mysql-test to be more robust. Fix that GRANT doesn't delete SSL options Change innobase_flush_log_at_trx_commit to uint. Don't rotate logs if we read a rotate log entry from the master.
2002-09-11 05:40:08 +02:00
inc_pos= event_len;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
break;
}
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
/*
If this event is originating from this server, don't queue it.
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
We don't check this for 3.23 events because it's simpler like this; 3.23
Fixes after merge
2003-10-08 11:01:58 +02:00
will be filtered anyway by the SQL slave thread which also tests the
server id (we must also keep this test in the SQL thread, in case somebody
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
upgrades a 4.0 slave which has a not-filtered relay log).
ANY event coming from ourselves can be ignored: it is obvious for queries;
for STOP_EVENT/ROTATE_EVENT/START_EVENT: these cannot come from ourselves
(--log-slave-updates would not log that) unless this slave is also its
direct master (an unsupported, useless setup!).
*/
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
pthread_mutex_lock(log_lock);
Replication: as the I/O thread filters on the server id, we must test replicate_same_server_id in it (on top of in the SQL thread).
2004-05-05 22:01:41 +02:00
if ((uint4korr(buf + SERVER_ID_OFFSET) == ::server_id) &&
BUG#23171: Illegal group log position Tail fixes after re-applying patches to older version of clone.
2007-03-22 08:32:41 +01:00
!mi->rli.replicate_same_server_id)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
/*
Do not write it to the relay log.
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
a) We still want to increment mi->master_log_pos, so that we won't
re-read this event from the master if the slave IO thread is now
stopped/restarted (more efficient if the events we are ignoring are big
LOAD DATA INFILE).
b) We want to record that we are skipping events, for the information of
the slave SQL thread, otherwise that thread may let
rli->group_relay_log_pos stay too small if the last binlog's event is
ignored.
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
But events which were generated by this slave and which do not exist in
the master's binlog (i.e. Format_desc, Rotate & Stop) should not increment
mi->master_log_pos.
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
*/
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
if (buf[EVENT_TYPE_OFFSET]!=FORMAT_DESCRIPTION_EVENT &&
buf[EVENT_TYPE_OFFSET]!=ROTATE_EVENT &&
buf[EVENT_TYPE_OFFSET]!=STOP_EVENT)
fixes after merge of 4.1.
2005-10-13 00:29:23 +02:00
{
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
mi->master_log_pos+= inc_pos;
fixes after merge of 4.1.
2005-10-13 00:29:23 +02:00
memcpy(rli->ign_master_log_name_end, mi->master_log_name, FN_REFLEN);
DBUG_ASSERT(rli->ign_master_log_name_end[0]);
rli->ign_master_log_pos_end= mi->master_log_pos;
}
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
rli->relay_log.signal_update(); // the slave SQL thread needs to re-check
Merge mysql.com:/home/my/mysql-5.0 into mysql.com:/home/my/mysql-5.1
2006-11-22 13:11:36 +01:00
DBUG_PRINT("info", ("master_log_pos: %lu, event originating from the same server, ignored",
Remove compiler warnings (Mostly in DBUG_PRINT() and unused arguments) Fixed bug in query cache when used with traceing (--with-debug) Fixed memory leak in mysqldump Removed warnings from mysqltest scripts (replaced -- with #)
2006-11-20 21:42:06 +01:00
(ulong) mi->master_log_pos));
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
}
Fixed many compiler warnings Fixed bugs in group_concat with ORDER BY and DISTINCT (Bugs #2695, #3381 and #3319) Fixed crash when doing rollback in slave and the io thread catched up with the sql thread Set locked_in_memory properly
2004-04-05 12:56:05 +02:00
else
{
/* write the event to the relay log */
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
if (likely(!(rli->relay_log.appendv(buf,event_len,0))))
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
{
mi->master_log_pos+= inc_pos;
Remove compiler warnings (Mostly in DBUG_PRINT() and unused arguments) Fixed bug in query cache when used with traceing (--with-debug) Fixed memory leak in mysqldump Removed warnings from mysqltest scripts (replaced -- with #)
2006-11-20 21:42:06 +01:00
DBUG_PRINT("info", ("master_log_pos: %lu", (ulong) mi->master_log_pos));
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
rli->relay_log.harvest_bytes_written(&rli->log_space_total);
}
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
else
Merge mysql.com:/home/mysql_src/mysql-4.1 into mysql.com:/home/mysql_src/mysql-5.0; a very bad automerge (issues with non-ascii chars), plus some hard conflicts I'll fix by hand in a next cset
2005-10-12 21:58:02 +02:00
error= 3;
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
rli->ign_master_log_name_end[0]= 0; // last event is not ignored
Fixed many compiler warnings Fixed bugs in group_concat with ORDER BY and DISTINCT (Bugs #2695, #3381 and #3319) Fixed crash when doing rollback in slave and the io thread catched up with the sql thread Set locked_in_memory properly
2004-04-05 12:56:05 +02:00
}
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
pthread_mutex_unlock(log_lock);
See each file's changeset for details. - Comments for future devs. - Start_log_event::exec_event() : when we hit it, do a rollback. - We don't need LOG_EVENT_FORCED_ROTATE_F. - Stop_log_event::exec_event() : when we hit it, we needn't clean anything. - Removed LOG_EVENT_TIME_F and LOG_EVENT_FORCED_ROTATE_F. - We don't need Stop events in the relay log. - Now filtering of server id is done in the I/O thread first.
2003-06-12 16:20:31 +02:00
err:
misc replication bugfixes including some needed modifications in IO_CACHE likely() and unlikely() branch prediction compiler hint macros clean-up of comments
2002-01-27 06:26:24 +01:00
pthread_mutex_unlock(&mi->data_lock);
After merge fixes: - Remove compiler warnings - Fix merge errors - Indentation fixes
2006-12-15 05:21:15 +01:00
DBUG_PRINT("info", ("error: %d", error));
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(error);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
void end_relay_log_info(Relay_log_info* rli)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
DBUG_ENTER("end_relay_log_info");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (!rli->inited)
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
DBUG_VOID_RETURN;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (rli->info_fd >= 0)
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
{
end_io_cache(&rli->info_file);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
(void) my_close(rli->info_fd, MYF(MY_WME));
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
rli->info_fd = -1;
}
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (rli->cur_log_fd >= 0)
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
{
end_io_cache(&rli->cache_buf);
(void)my_close(rli->cur_log_fd, MYF(MY_WME));
rli->cur_log_fd = -1;
}
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
rli->inited = 0;
Better fix for bug #791: At binlog rotation, INSERTs may not find their way into the binlog
2003-07-14 13:59:26 +02:00
rli->relay_log.close(LOG_CLOSE_INDEX | LOG_CLOSE_STOP_EVENT);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
rli->relay_log.harvest_bytes_written(&rli->log_space_total);
4 small items in this: - when we don't have in_addr_t, use uint32. - a forgotten initialization of slave_proxy_id in sql/log_event.cc (was not really "forgot", was "we needn't init it there", but there was one case where we needed...). - made slave_proxy_id always meaningful in THD and Log_event, so we can rely more on it (no need to test if it's meaningful). THD::slave_proxy_id is equal to THD::thread_id except for the slave SQL thread. - clean up the slave's temporary table (i.e. free their memory) when slave server shuts down.
2003-10-31 23:20:23 +01:00
/*
Delete the slave's temporary tables from memory.
In the future there will be other actions than this, to ensure persistance
of slave's temp tables after shutdown.
*/
rli->close_temporary_tables();
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
DBUG_VOID_RETURN;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Bug#46013: rpl_extraColmaster_myisam fails on pb2 Bug#45243: crash on win in sql thread clear_tables_to_lock() -> free() Bug#45242: crash on win in mysql_close() -> free() Bug#45238: rpl_slave_skip, rpl_change_master failed (lost connection) for STOP SLAVE Bug#46030: rpl_truncate_3innodb causes server crash on windows Bug#46014: rpl_stm_reset_slave crashes the server sporadically in pb2 When killing a user session on the server, it's necessary to interrupt (notify) the thread associated with the session that the connection is being killed so that the thread is woken up if waiting for I/O. On a few platforms (Mac, Windows and HP-UX) where the SIGNAL_WITH_VIO_CLOSE flag is defined, this interruption procedure is to asynchronously close the underlying socket of the connection. In order to enable this schema, each connection serving thread registers its VIO (I/O interface) so that other threads can access it and close the connection. But only the owner thread of the VIO might delete it as to guarantee that other threads won't see freed memory (the thread unregisters the VIO before deleting it). A side note: closing the socket introduces a harmless race that might cause a thread attempt to read from a closed socket, but this is deemed acceptable. The problem is that this infrastructure was meant to only be used by server threads, but the slave I/O thread was registering the VIO of a mysql handle (a client API structure that represents a connection to another server instance) as a active connection of the thread. But under some circumstances such as network failures, the client API might destroy the VIO associated with a handle at will, yet the VIO wouldn't be properly unregistered. This could lead to accesses to freed data if a thread attempted to kill a slave I/O thread whose connection was already broken. There was a attempt to work around this by checking whether the socket was being interrupted, but this hack didn't work as intended due to the aforementioned race -- attempting to read from the socket would yield a "bad file descriptor" error. The solution is to add a hook to the client API that is called from the client code before the VIO of a handle is deleted. This hook allows the slave I/O thread to detach the active vio so it does not point to freed memory.
2009-08-13 22:07:20 +02:00
/**
Hook to detach the active VIO before closing a connection handle.
The client API might close the connection (and associated data)
in case it encounters a unrecoverable (network) error. This hook
is called from the client code before the VIO handle is deleted
allows the thread to detach the active vio so it does not point
to freed memory.
Other calls to THD::clear_active_vio throughout this module are
redundant due to the hook but are left in place for illustrative
purposes.
*/
extern "C" void slave_io_thread_detach_vio()
{
#ifdef SIGNAL_WITH_VIO_CLOSE
THD *thd= current_thd;
Bug#47525: MySQL crashed (Federated) On Mac OS X or Windows, sending a SIGHUP to the server or a asynchronous flush (triggered by flush_time), would cause the server to crash. The problem was that a hook used to detach client API handles wasn't prepared to handle cases where the thread does not have a associated session. The solution is to verify whether the thread has a associated session before trying to detach a handle.
2009-09-30 23:38:02 +02:00
if (thd && thd->slave_thread)
Bug#46013: rpl_extraColmaster_myisam fails on pb2 Bug#45243: crash on win in sql thread clear_tables_to_lock() -> free() Bug#45242: crash on win in mysql_close() -> free() Bug#45238: rpl_slave_skip, rpl_change_master failed (lost connection) for STOP SLAVE Bug#46030: rpl_truncate_3innodb causes server crash on windows Bug#46014: rpl_stm_reset_slave crashes the server sporadically in pb2 When killing a user session on the server, it's necessary to interrupt (notify) the thread associated with the session that the connection is being killed so that the thread is woken up if waiting for I/O. On a few platforms (Mac, Windows and HP-UX) where the SIGNAL_WITH_VIO_CLOSE flag is defined, this interruption procedure is to asynchronously close the underlying socket of the connection. In order to enable this schema, each connection serving thread registers its VIO (I/O interface) so that other threads can access it and close the connection. But only the owner thread of the VIO might delete it as to guarantee that other threads won't see freed memory (the thread unregisters the VIO before deleting it). A side note: closing the socket introduces a harmless race that might cause a thread attempt to read from a closed socket, but this is deemed acceptable. The problem is that this infrastructure was meant to only be used by server threads, but the slave I/O thread was registering the VIO of a mysql handle (a client API structure that represents a connection to another server instance) as a active connection of the thread. But under some circumstances such as network failures, the client API might destroy the VIO associated with a handle at will, yet the VIO wouldn't be properly unregistered. This could lead to accesses to freed data if a thread attempted to kill a slave I/O thread whose connection was already broken. There was a attempt to work around this by checking whether the socket was being interrupted, but this hack didn't work as intended due to the aforementioned race -- attempting to read from the socket would yield a "bad file descriptor" error. The solution is to add a hook to the client API that is called from the client code before the VIO of a handle is deleted. This hook allows the slave I/O thread to detach the active vio so it does not point to freed memory.
2009-08-13 22:07:20 +02:00
thd->clear_active_vio();
#endif
}
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
/*
Try to connect until successful or slave killed
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
SYNPOSIS
safe_connect()
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
thd Thread handler for slave
mysql MySQL connection handle
mi Replication handle
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
RETURN
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
0 ok
# Error
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
*/
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int safe_connect(THD* thd, MYSQL* mysql, Master_info* mi)
Import changeset
2000-07-31 21:29:14 +02:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("safe_connect");
DBUG_RETURN(connect_to_master(thd, mysql, mi, 0, 0));
Import changeset
2000-07-31 21:29:14 +02:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
/*
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
SYNPOSIS
connect_to_master()
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
IMPLEMENTATION
Try to connect until successful or slave killed or we have retried
master_retry_count times
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
*/
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int connect_to_master(THD* thd, MYSQL* mysql, Master_info* mi,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
bool reconnect, bool suppress_warnings)
Import changeset
2000-07-31 21:29:14 +02:00
{
added loops and other cool stuff to mysqltest fixed slave to preserve temp tables on slave stop; slave start added abort-slave-event-count option to mysqld to test the above added a test case for temp tables with a contantly aborting slave removed warning in sql_parse.cc fixed thimble.cc to compile
2000-12-02 18:11:50 +01:00
int slave_was_killed;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
int last_errno= -2; // impossible error
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
ulong err_count=0;
Fixed test cases for innobase Fixed bug introduced with last ORDER BY optimization Changed log position to longlong to avoid warnings.
2001-03-07 22:50:44 +01:00
char llbuff[22];
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
DBUG_ENTER("connect_to_master");
Fixed test cases for innobase Fixed bug introduced with last ORDER BY optimization Changed log position to longlong to avoid warnings.
2001-03-07 22:50:44 +01:00
mysql-test/mysql-test-run cleanup - remove unneeded daeamon restarts sql/log_event.cc fixed bug - do not confuse confused slaves even further if they try to read less than LOG_EVENT_HEADER_LEN before the end of file sql/mysqlbinlog.cc give error if constructor fails sql/mysqld.cc --disconnect-slave-event-count for replication debugging sql/slave.cc --disconnect-slave-event-count cleanup to remove unneeded sleeps when things go wrong - sleep only when we have to sql/slave.h --disconnect-slave-event-count test cases for ALTER TABLE ORDER BY and replication offset off by 22 confusion mystery bug (both pass)
2000-11-22 08:23:31 +01:00
#ifndef DBUG_OFF
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
mi->events_till_disconnect = disconnect_slave_event_count;
mysql-test/mysql-test-run cleanup - remove unneeded daeamon restarts sql/log_event.cc fixed bug - do not confuse confused slaves even further if they try to read less than LOG_EVENT_HEADER_LEN before the end of file sql/mysqlbinlog.cc give error if constructor fails sql/mysqld.cc --disconnect-slave-event-count for replication debugging sql/slave.cc --disconnect-slave-event-count cleanup to remove unneeded sleeps when things go wrong - sleep only when we have to sql/slave.h --disconnect-slave-event-count test cases for ALTER TABLE ORDER BY and replication offset off by 22 confusion mystery bug (both pass)
2000-11-22 08:23:31 +01:00
#endif
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
ulong client_flag= CLIENT_REMEMBER_OPTIONS;
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
if (opt_slave_compressed_protocol)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
client_flag=CLIENT_COMPRESS; /* We will use compression */
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
mysql_options(mysql, MYSQL_OPT_CONNECT_TIMEOUT, (char *) &slave_net_timeout);
mysql_options(mysql, MYSQL_OPT_READ_TIMEOUT, (char *) &slave_net_timeout);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
#ifdef HAVE_OPENSSL
if (mi->ssl)
Bug#19991 CHANGE MASTER need option ssl-verify-server-cert - Add MASTER_SSL_VERIFY_SERVER_CERT option to CHANGE MASTER TO - Add Master_Ssl_Serify_Server_Cert to SHOW SLAVE STATUS - Save and restore ssl_verify_server_cert to master info file setting it to disabled as default.
2007-03-29 15:09:57 +02:00
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mysql_ssl_set(mysql,
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
mi->ssl_key[0]?mi->ssl_key:0,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mi->ssl_cert[0]?mi->ssl_cert:0,
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
mi->ssl_ca[0]?mi->ssl_ca:0,
mi->ssl_capath[0]?mi->ssl_capath:0,
mi->ssl_cipher[0]?mi->ssl_cipher:0);
Bug#19991 CHANGE MASTER need option ssl-verify-server-cert - Add MASTER_SSL_VERIFY_SERVER_CERT option to CHANGE MASTER TO - Add Master_Ssl_Serify_Server_Cert to SHOW SLAVE STATUS - Save and restore ssl_verify_server_cert to master info file setting it to disabled as default.
2007-03-29 15:09:57 +02:00
mysql_options(mysql, MYSQL_OPT_SSL_VERIFY_SERVER_CERT,
&mi->ssl_verify_server_cert);
}
Implemented replication over SSL Added proper options to CHANGE MASTER TO, new fields to SHOW SLAVE STATUS, Honoring this parameters during connection to master. Introduced new format of master.info file
2003-09-01 13:16:20 +02:00
#endif
Cleanup after split of libmysql.c to client.c and libmysql.c. A 4.1 master/slave will now use the 4.1 protocol Fixed wrong value for SQLSTATE_LENGTH Added CLIENT_REMEMBER_OPTIONS to mysql_real_connect() Changed mysql_port and mysql_unix_port to mysqld_xxxx
2003-06-14 10:37:42 +02:00
mysql_options(mysql, MYSQL_SET_CHARSET_NAME, default_charset_info->csname);
/* This one is not strictly needed but we have it here for completeness */
mysql_options(mysql, MYSQL_SET_CHARSET_DIR, (char *) charsets_dir);
fixed replication assert failure ( found by Jeremy Zawodny)
2002-03-08 23:02:11 +01:00
while (!(slave_was_killed = io_slave_killed(thd,mi)) &&
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
(reconnect ? mysql_reconnect(mysql) != 0 :
mysql_real_connect(mysql, mi->host, mi->user, mi->password, 0,
mi->port, 0, client_flag) == 0))
Import changeset
2000-07-31 21:29:14 +02:00
{
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
/* Don't repeat last error */
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
if ((int)mysql_errno(mysql) != last_errno)
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
{
SCRUM common parts of libmysql.c and sql/mini_client.cc moved to sql-common/client.c names of functions now are same in both cases
2003-05-02 18:07:41 +02:00
last_errno=mysql_errno(mysql);
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
suppress_warnings= 0;
BUG#24954 (Last_errno and Last_error not set after master_retry_count was reached): Adding new fields Last_{IO,SQL}_Errno and Last_{IO,SQL}_Error to output of SHOW SLAVE STATUS to hold errors from I/O and SQL thread respectively. Old fields Last_Error and Last_Errno are aliases for Last_SQL_Error and Last_SQL_Errno respectively. Fields are added last to output of SHOW SLAVE STATUS to allow old applications to use the same positional arguments into the row, while allowing new application to benefit from the added information. In addition, some new error codes are added (especially for the I/O thread) to be able to provide sensible error message.
2007-06-09 07:19:37 +02:00
mi->report(ERROR_LEVEL, last_errno,
"error %s to master '%s@%s:%d'"
" - retry-time: %d retries: %lu",
(reconnect ? "reconnecting" : "connecting"),
mi->user, mi->host, mi->port,
mi->connect_retry, master_retry_count);
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
}
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
/*
By default we try forever. The reason is that failure will trigger
master election, so if the user did not set master_retry_count we
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
do not want to have election triggered on the first failure to
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
connect
fixes for mysql-test to run without manager a bit more work on failsafe replication
2001-10-11 21:54:06 +02:00
*/
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
if (++err_count == master_retry_count)
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
{
slave_was_killed=1;
cleanup removal of duplicate code in mf_iocache.cc work on failsafe replication work on SEQ_READ_APPEND io cache
2001-10-23 21:28:03 +02:00
if (reconnect)
change_rpl_status(RPL_ACTIVE_SLAVE,RPL_LOST_SOLDIER);
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
break;
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
safe_sleep(thd,mi->connect_retry,(CHECK_KILLED_FUNC)io_slave_killed,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
(void*)mi);
Import changeset
2000-07-31 21:29:14 +02:00
}
added loops and other cool stuff to mysqltest fixed slave to preserve temp tables on slave stop; slave start added abort-slave-event-count option to mysqld to test the above added a test case for temp tables with a contantly aborting slave removed warning in sql_parse.cc fixed thimble.cc to compile
2000-12-02 18:11:50 +01:00
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
if (!slave_was_killed)
{
BUG#44270: RESET SLAVE does not reset Last_IO_Error or Last_IO_Errno The server was not cleaning the last IO error and error number when resetting slave. This patch addresses this issue by backporting into 5.1 part of the patch in BUG 34654. A fix for this issue had already been pushed into 6.0 as part of the aforementioned bug, however the patch also included some refactoring. The fix for 5.1 does not take into account the refactoring part.
2009-06-03 16:14:18 +02:00
mi->clear_error(); // clear possible left over reconnect error
cleanup removal of duplicate code in mf_iocache.cc work on failsafe replication work on SEQ_READ_APPEND io cache
2001-10-23 21:28:03 +02:00
if (reconnect)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
{
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
if (!suppress_warnings && global_system_variables.log_warnings)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
sql_print_information("Slave: connected to master '%s@%s:%d',\
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
replication resumed in log '%s' at position %s", mi->user,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
mi->host, mi->port,
IO_RPL_LOG_NAME,
llstr(mi->master_log_pos,llbuff));
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
}
cleanup removal of duplicate code in mf_iocache.cc work on failsafe replication work on SEQ_READ_APPEND io cache
2001-10-23 21:28:03 +02:00
else
{
change_rpl_status(RPL_IDLE_SLAVE,RPL_ACTIVE_SLAVE);
WL1019: complete patch. Reapplied patch to the clean tree to get rid of multiple typos in CS comments and unify the patch.
2006-01-19 03:56:06 +01:00
general_log_print(thd, COM_CONNECT_OUT, "%s@%s:%d",
mi->user, mi->host, mi->port);
cleanup removal of duplicate code in mf_iocache.cc work on failsafe replication work on SEQ_READ_APPEND io cache
2001-10-23 21:28:03 +02:00
}
changed signal by fd close to signal by vio_close added support for kill expr fixed coredump in set @a := foo; added testcase for user_var added testcase for kill
2001-03-14 07:07:12 +01:00
#ifdef SIGNAL_WITH_VIO_CLOSE
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
thd->set_active_vio(mysql->net.vio);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
#endif
Added usage of --master-retry-count to make mysql-test-run safer.
2001-10-03 15:27:20 +02:00
}
Changing the default of libmysqlclient : it's now NO reconnection. All our programs which use mysql_real_connect() and mysql_connect() are updated accordingly, though I have deliberately made mysqlimport not reconnect anymore (already true for mysqldump >= 4.1.8). All Connector devs have been warned about the change I'm doing here - which was agreed with Monty, and fixes BUG#2555.
2004-12-09 14:44:10 +01:00
mysql->reconnect= 1;
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
DBUG_PRINT("exit",("slave_was_killed: %d", slave_was_killed));
DBUG_RETURN(slave_was_killed);
Import changeset
2000-07-31 21:29:14 +02:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
cleanup removal of duplicate code in mf_iocache.cc work on failsafe replication work on SEQ_READ_APPEND io cache
2001-10-23 21:28:03 +02:00
/*
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
safe_reconnect()
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
Merge with 4.0.11
2003-02-04 20:52:14 +01:00
IMPLEMENTATION
Try to connect until successful or slave killed or we have retried
master_retry_count times
cleanup removal of duplicate code in mf_iocache.cc work on failsafe replication work on SEQ_READ_APPEND io cache
2001-10-23 21:28:03 +02:00
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
static int safe_reconnect(THD* thd, MYSQL* mysql, Master_info* mi,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
bool suppress_warnings)
cleanup removal of duplicate code in mf_iocache.cc work on failsafe replication work on SEQ_READ_APPEND io cache
2001-10-23 21:28:03 +02:00
{
Fixed some new memory leaks Updated VC++ files
2003-06-24 11:10:35 +02:00
DBUG_ENTER("safe_reconnect");
DBUG_RETURN(connect_to_master(thd, mysql, mi, 1, suppress_warnings));
cleanup removal of duplicate code in mf_iocache.cc work on failsafe replication work on SEQ_READ_APPEND io cache
2001-10-23 21:28:03 +02:00
}
New SET syntax & system variables. Made a some new buffers thread specific and changeable. Resize of key_buffer. AUTO_COMMIT -> AUTOCOMMIT Fixed mutex bug in DROP DATABASE Fixed bug when using auto_increment as second part of a key where first part could include NULL. Split handler->extra() to extra() and extra_opt() to be able to support thread specific buffers. Don't write message to error log when slave reconnects becasue of timeout. Fixed possible update problem when using DELETE/UPDATE on small tables (In some cases we used index even if table scanning would be better) A lot of minior code cleanups
2002-07-23 17:31:22 +02:00
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
/*
Store the file and position where the execute-slave thread are in the
relay log.
SYNOPSIS
flush_relay_log_info()
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
rli Relay log information
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
NOTES
- As this is only called by the slave thread, we don't need to
have a lock on this.
- If there is an active transaction, then we don't update the position
in the relay log. This is to ensure that we re-execute statements
if we die in the middle of an transaction that was rolled back.
- As a transaction never spans binary logs, we don't have to handle the
case where we do a relay-log-rotation in the middle of the transaction.
If this would not be the case, we would have to ensure that we
don't delete the relay log file where the transaction started when
we switch to a new relay log file.
TODO
- Change the log file information to a binary format to avoid calling
longlong2str.
RETURN VALUES
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
0 ok
1 write error
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
*/
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
bool flush_relay_log_info(Relay_log_info* rli)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
bool error=0;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("flush_relay_log_info");
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
if (unlikely(rli->no_storage))
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(0);
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
IO_CACHE *file = &rli->info_file;
char buff[FN_REFLEN*2+22*2+4], *pos;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
my_b_seek(file, 0L);
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
pos=strmov(buff, rli->group_relay_log_name);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
*pos++='\n';
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
pos=longlong2str(rli->group_relay_log_pos, pos, 10);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
*pos++='\n';
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
pos=strmov(pos, rli->group_master_log_name);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
*pos++='\n';
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
pos=longlong2str(rli->group_master_log_pos, pos, 10);
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
*pos='\n';
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 11:59:39 +02:00
if (my_b_write(file, (uchar*) buff, (size_t) (pos-buff)+1))
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
error=1;
if (flush_io_cache(file))
error=1;
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
Undoing 2 parts of changeset 1.1730.1.1 : - the one about BUG#2921 - the one about relay log flushing Both will be rewritten in a next changeset (this one will not be pushed before the next changeset).
2004-03-10 16:30:47 +01:00
/* Flushing the relay log is done by the slave I/O thread */
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_RETURN(error);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
Added comment blocks for all functions
2002-10-29 23:12:47 +01:00
Called when we notice that the current "hot" log got rotated under our feet.
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
*/
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
static IO_CACHE *reopen_relay_log(Relay_log_info *rli, const char **errmsg)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
DBUG_ENTER("reopen_relay_log");
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(rli->cur_log != &rli->cache_buf);
DBUG_ASSERT(rli->cur_log_fd == -1);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
IO_CACHE *cur_log = rli->cur_log=&rli->cache_buf;
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
if ((rli->cur_log_fd=open_binlog(cur_log,rli->event_relay_log_name,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
errmsg)) <0)
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(0);
Ensure that BEGIN / COMMIT is handled properly if slave dies Added syntax support for CREATE TABLE foo (a char CHARACTER SET latin1) CHARSET=latin1;
2002-12-11 14:46:39 +01:00
/*
We want to start exactly where we was before:
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
relay_log_pos Current log pos
pending Number of bytes already processed from the event
Ensure that BEGIN / COMMIT is handled properly if slave dies Added syntax support for CREATE TABLE foo (a char CHARACTER SET latin1) CHARSET=latin1;
2002-12-11 14:46:39 +01:00
*/
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
rli->event_relay_log_pos= max(rli->event_relay_log_pos, BIN_LOG_HEADER_SIZE);
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
my_b_seek(cur_log,rli->event_relay_log_pos);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(cur_log);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
BUG#41924: high-level replication functions are not commented Adding comments to some of the high-level functions in replication.
2009-01-09 13:49:24 +01:00
/**
Reads next event from the relay log. Should be called from the
slave IO thread.
@param rli Relay_log_info structure for the slave IO thread.
@return The event read, or NULL on error. If an error occurs, the
error is reported through the sql_print_information() or
sql_print_error() functions.
*/
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
static Log_event* next_event(Relay_log_info* rli)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
Log_event* ev;
IO_CACHE* cur_log = rli->cur_log;
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
pthread_mutex_t *log_lock = rli->relay_log.get_log_lock();
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
const char* errmsg=0;
THD* thd = rli->sql_thd;
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_ENTER("next_event");
sql/Makefile.am: better mysqld dependencies sql/discover.cc: close(-1) bug sql/slave.cc: DBUG_ENTER/RETURN tags ha_archive.cc: double-close bug
2006-05-30 18:21:03 +02:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
DBUG_ASSERT(thd != 0);
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
#ifndef DBUG_OFF
if (abort_slave_event_count && !rli->events_till_abort--)
DBUG_RETURN(0);
#endif
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
/*
For most operations we need to protect rli members with data_lock,
Implemented support for START SLAVE UNTIL (WL#868)
2003-09-13 22:13:41 +02:00
so we assume calling function acquired this mutex for us and we will
hold it for the most of the loop below However, we will release it
whenever it is worth the hassle, and in the cases when we go into a
pthread_cond_wait() with the non-data_lock mutex
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
*/
Implemented support for START SLAVE UNTIL (WL#868)
2003-09-13 22:13:41 +02:00
safe_mutex_assert_owner(&rli->data_lock);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Changed relay_log_space to ulonglong Removed compiler warnings.
2002-08-23 14:14:01 +02:00
while (!sql_slave_killed(thd,rli))
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
{
/*
We can have two kinds of log reading:
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
hot_log:
rli->cur_log points at the IO_CACHE of relay_log, which
is actively being updated by the I/O thread. We need to be careful
in this case and make sure that we are not looking at a stale log that
has already been rotated. If it has been, we reopen the log.
The other case is much simpler:
We just have a read only log that nobody else will be updating.
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
*/
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
bool hot_log;
if ((hot_log = (cur_log != &rli->cache_buf)))
{
DBUG_ASSERT(rli->cur_log_fd == -1); // foreign descriptor
pthread_mutex_lock(log_lock);
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
/*
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Reading xxx_file_id is safe because the log will only
be rotated when we hold relay_log.LOCK_log
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
*/
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
if (rli->relay_log.get_open_count() != rli->cur_log_old_open_count)
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
// The master has switched to a new log file; Reopen the old log file
cur_log=reopen_relay_log(rli, &errmsg);
pthread_mutex_unlock(log_lock);
if (!cur_log) // No more log files
goto err;
hot_log=0; // Using old binary log
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
}
Fix for bug #10798: If relay log fails to rotate, slave will crash after The relay log may not be open for some reason (e.g. disk error) after rotation, and using it causes the slave crash. Fix: check we have it open before access, return error otherwise.
2006-12-18 14:38:59 +01:00
/*
As there is no guarantee that the relay is open (for example, an I/O
error during a write by the slave I/O thread may have closed it), we
have to test it.
*/
if (!my_b_inited(cur_log))
goto err;
I committed the same changeset in my tree yesterday, but broke the tree since, so I commit again in a fresh tree. Fix for bug#763 (Relay_log_space too big by 4 bytes), plus comments and DBUG_PRINT, and we don't start replication if --bootstrap.
2003-07-03 01:08:36 +02:00
#ifndef DBUG_OFF
{
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
/* This is an assertion which sometimes fails, let's try to track it */
I committed the same changeset in my tree yesterday, but broke the tree since, so I commit again in a fresh tree. Fix for bug#763 (Relay_log_space too big by 4 bytes), plus comments and DBUG_PRINT, and we don't start replication if --bootstrap.
2003-07-03 01:08:36 +02:00
char llbuf1[22], llbuf2[22];
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
DBUG_PRINT("info", ("my_b_tell(cur_log)=%s rli->event_relay_log_pos=%s",
open binlog index file *before* tc_log->open() and binlog itself *after*
2005-02-01 15:36:48 +01:00
llstr(my_b_tell(cur_log),llbuf1),
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
llstr(rli->event_relay_log_pos,llbuf2)));
DBUG_ASSERT(my_b_tell(cur_log) >= BIN_LOG_HEADER_SIZE);
DBUG_ASSERT(my_b_tell(cur_log) == rli->event_relay_log_pos);
I committed the same changeset in my tree yesterday, but broke the tree since, so I commit again in a fresh tree. Fix for bug#763 (Relay_log_space too big by 4 bytes), plus comments and DBUG_PRINT, and we don't start replication if --bootstrap.
2003-07-03 01:08:36 +02:00
}
#endif
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
Relay log is always in new format - if the master is 3.23, the
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
I/O thread will convert the format for us.
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
A problem: the description event may be in a previous relay log. So if
the slave has been shutdown meanwhile, we would have to look in old relay
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
logs, which may even have been deleted. So we need to write this
description event at the beginning of the relay log.
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
When the relay log is created when the I/O thread starts, easy: the
master will send the description event and we will queue it.
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
But if the relay log is created by new_file(): then the solution is:
WL#3153 "Split logs". Recommit with post-review fixes
2006-05-05 08:45:58 +02:00
MYSQL_BIN_LOG::open() will write the buffered description event.
fixes for slave backward compat fixed bug in LOAD DATA FROM MASTER fixed rpl000001.result Slave now replicates 3.23 master, with the exception of LOAD DATA INFILE, which is still buggy. Will push this one after the pull/merge
2002-02-07 23:29:46 +01:00
*/
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
if ((ev=Log_event::read_log_event(cur_log,0,
rli->relay_log.description_event_for_exec)))
open binlog index file *before* tc_log->open() and binlog itself *after*
2005-02-01 15:36:48 +01:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
DBUG_ASSERT(thd==rli->sql_thd);
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
/*
read it while we have a lock, to avoid a mutex lock in
inc_event_relay_log_pos()
*/
rli->future_event_relay_log_pos= my_b_tell(cur_log);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (hot_log)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
pthread_mutex_unlock(log_lock);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(ev);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
DBUG_ASSERT(thd==rli->sql_thd);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
if (opt_reckless_slave) // For mysql-test
reckless slave option bison 1.34 yacc rule compliance more robust handling of relay log corruption
2002-04-16 01:09:30 +02:00
cur_log->error = 0;
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
if (cur_log->error < 0)
reckless slave option bison 1.34 yacc rule compliance more robust handling of relay log corruption
2002-04-16 01:09:30 +02:00
{
errmsg = "slave SQL thread aborted because of I/O error";
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
if (hot_log)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
pthread_mutex_unlock(log_lock);
reckless slave option bison 1.34 yacc rule compliance more robust handling of relay log corruption
2002-04-16 01:09:30 +02:00
goto err;
}
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (!cur_log->error) /* EOF */
{
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
/*
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
On a hot log, EOF means that there are no more updates to
process and we must block until I/O thread adds some and
signals us to continue
New error messages Test of unsigned BIGINT values Fixes for queries-per-hour Cleanup of replication code (comments and portability fixes) Make most of the binary log code 4G clean Changed syntax for GRANT ... QUERIES PER HOUR
2002-01-29 17:32:16 +01:00
*/
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
if (hot_log)
{
Very minor fixes for Seconds_Behind_Master column of SHOW SLAVE STATUS.
2004-12-16 22:38:42 +01:00
/*
We say in Seconds_Behind_Master that we have "caught up". Note that
for example if network link is broken but I/O slave thread hasn't
noticed it (slave_net_timeout not elapsed), then we'll say "caught
up" whereas we're not really caught up. Fixing that would require
internally cutting timeout in smaller pieces in network read, no
thanks. Another example: SQL has caught up on I/O, now I/O has read
a new event and is queuing it; the false "0" will exist until SQL
finishes executing the new event; it will be look abnormal only if
the events have old timestamps (then you get "many", 0, "many").
Bug #29309 Incorrect "Seconds_Behind_Master" value in SHOW SLAVE STATUS after FLUSH LOGS Report claims that Seconds_behind_master behaves unexpectedly. Code analysis shows that there is an evident flaw in that treating of FormatDescription event is wrong so that after FLUSH LOGS on slave the Seconds_behind_master's calculation slips and incorrect value can be reported to SHOW SLAVE STATUS. Even worse is that the gap between the correct and incorrect deltas grows with time. Fixed with prohibiting changes to rpl->last_master_timestamp by artifical events (any kind of). suggestion as comments is added how to fight with lack of info on the slave side by means of new heartbeat feature coming. The test can not be done ealily fully determistic.
2007-10-04 17:46:31 +02:00
Transient phases like this can be fixed with implemeting
Heartbeat event which provides the slave the status of the
master at time the master does not have any new update to send.
Seconds_Behind_Master would be zero only when master has no
more updates in binlog for slave. The heartbeat can be sent
in a (small) fraction of slave_net_timeout. Until it's done
rli->last_master_timestamp is temporarely (for time of
waiting for the following event) reset whenever EOF is
reached.
Very minor fixes for Seconds_Behind_Master column of SHOW SLAVE STATUS.
2004-12-16 22:38:42 +01:00
*/
Reverting to use time_t instead of my_time_t since the latter caused compile warnings on windows machines.
2007-04-17 08:39:38 +02:00
time_t save_timestamp= rli->last_master_timestamp;
Very minor fixes for Seconds_Behind_Master column of SHOW SLAVE STATUS.
2004-12-16 22:38:42 +01:00
rli->last_master_timestamp= 0;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
DBUG_ASSERT(rli->relay_log.get_open_count() ==
Fixes during review of new code
2005-10-13 18:24:01 +02:00
rli->cur_log_old_open_count);
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
if (rli->ign_master_log_name_end[0])
{
/* We generate and return a Rotate, to make our positions advance */
DBUG_PRINT("info",("seeing an ignored end segment"));
WL#1012: All changes as one single changeset. This includes both code and test cases.
2005-12-22 06:39:02 +01:00
ev= new Rotate_log_event(rli->ign_master_log_name_end,
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
0, rli->ign_master_log_pos_end,
fixes after merge of 4.1.
2005-10-13 00:29:23 +02:00
Rotate_log_event::DUP_NAME);
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
rli->ign_master_log_name_end[0]= 0;
Fixes during review of new code
2005-10-13 18:24:01 +02:00
pthread_mutex_unlock(log_lock);
Fix for BUG#13023: "SQL Thread is up but doesn't move forward". Details in slave.cc; in short we now record whenever the slave I/O thread ignores a master's event because of its server id, and use this info in the slave SQL thread to advance Exec_master_log_pos. Because if we do not, this variable stays at the position of the last executed event, i.e. the last *non-ignored* executed one, which may not be the last of the master's binlog (and so the slave *looks* behind the master though it's data-wise it's not).
2005-10-12 13:29:55 +02:00
if (unlikely(!ev))
{
errmsg= "Slave SQL thread failed to create a Rotate event "
"(out of memory?), SHOW SLAVE STATUS may be inaccurate";
goto err;
}
ev->server_id= 0; // don't be ignored by slave SQL thread
DBUG_RETURN(ev);
}
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
/*
We can, and should release data_lock while we are waiting for
update. If we do not, show slave status will block
*/
pthread_mutex_unlock(&rli->data_lock);
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
/*
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Possible deadlock :
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
- the I/O thread has reached log_space_limit
- the SQL thread has read all relay logs, but cannot purge for some
reason:
* it has already purged all logs except the current one
* there are other logs than the current one but they're involved in
a transaction that finishes in the current one (or is not finished)
Solution :
Wake up the possibly waiting I/O thread, and set a boolean asking
the I/O thread to temporarily ignore the log_space_limit
constraint, because we do not want the I/O thread to block because of
space (it's ok if it blocks for any other reason (e.g. because the
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
master does not send anything). Then the I/O thread stops waiting
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
and reads more events.
The SQL thread decides when the I/O thread should take log_space_limit
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
into account again : ignore_log_space_limit is reset to 0
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
in purge_first_log (when the SQL thread purges the just-read relay
log), and also when the SQL thread starts. We should also reset
ignore_log_space_limit to 0 when the user does RESET SLAVE, but in
fact, no need as RESET SLAVE requires that the slave
Simplified a test. thd->enter_cond() and exit_cond(), so that the I/O thread accepts to stop when it's waiting for relay log space. Reset ignore_log_space_limit to 0 when the SQL thread terminates.
2003-06-15 12:01:51 +02:00
be stopped, and the SQL thread sets ignore_log_space_limit to 0 when
it stops.
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
*/
pthread_mutex_lock(&rli->log_space_lock);
// prevent the I/O thread from blocking next times
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
rli->ignore_log_space_limit= 1;
- Fix for memory leak if the SQL slave thread is killed just after reading an event. - A few more mutex locks and unlocks of rli.log_space_lock for doing clean reads of rli.ignore_log_space_limit - Broadcast after unlock, not before (small speed optimisation).
2003-05-25 23:09:46 +02:00
/*
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
If the I/O thread is blocked, unblock it. Ok to broadcast
after unlock, because the mutex is only destroyed in
~Relay_log_info(), i.e. when rli is destroyed, and rli will
not be destroyed before we exit the present function.
- Fix for memory leak if the SQL slave thread is killed just after reading an event. - A few more mutex locks and unlocks of rli.log_space_lock for doing clean reads of rli.ignore_log_space_limit - Broadcast after unlock, not before (small speed optimisation).
2003-05-25 23:09:46 +02:00
*/
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
pthread_mutex_unlock(&rli->log_space_lock);
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
pthread_cond_broadcast(&rli->log_space_cond);
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
// Note that wait_for_update unlocks lock_log !
Clearer states in SHOW PROCESSLIST for replication threads. For example the Binlog_dump thread (on the master) sometimes showed "Slave:". And there were confusing messages where "binlog" was employed instead of "relay log".
2003-08-25 14:13:58 +02:00
rli->relay_log.wait_for_update(rli->sql_thd, 1);
New variable rli->ignore_log_space_limit to resolve a deadlock between I/O and SQL threads in replication when relay_log_space is too small. This fixes bug #79.
2003-03-17 22:51:56 +01:00
// re-acquire data lock since we released it earlier
pthread_mutex_lock(&rli->data_lock);
A change of behaviour of Seconds_Behind_Master from SHOW SLAVE STATUS. It's going into 4.1 because old behaviour was somewhat nonsensical (kind of bug). Changes are that if repl threads are down or disconnected the column will be NULL, and if master is idle the column will not grow indefinitely anymore.
2004-12-16 18:12:22 +01:00
rli->last_master_timestamp= save_timestamp;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
continue;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
If the log was not hot, we need to move to the next log in
sequence. The next log could be hot or cold, we deal with both
cases separately after doing some common initialization
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
*/
end_io_cache(cur_log);
DBUG_ASSERT(rli->cur_log_fd >= 0);
my_close(rli->cur_log_fd, MYF(MY_WME));
rli->cur_log_fd = -1;
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
if (relay_log_purge)
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
/*
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
purge_first_log will properly set up relay log coordinates in rli.
If the group's coordinates are equal to the event's coordinates
(i.e. the relay log was not rotated in the middle of a group),
we can purge this relay log too.
We do ulonglong and string comparisons, this may be slow but
- purging the last relay log is nice (it can save 1GB of disk), so we
like to detect the case where we can do it, and given this,
- I see no better detection method
- purge_first_log is not called that often
*/
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
if (rli->relay_log.purge_first_log
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
(rli,
rli->group_relay_log_pos == rli->event_relay_log_pos
&& !strcmp(rli->group_relay_log_name,rli->event_relay_log_name)))
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
{
errmsg = "Error purging processed logs";
goto err;
}
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
}
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
else
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
/*
If hot_log is set, then we already have a lock on
LOCK_log. If not, we have to get the lock.
According to Sasha, the only time this code will ever be executed
is if we are recovering from a bug.
*/
if (rli->relay_log.find_next_log(&rli->linfo, !hot_log))
{
errmsg = "error switching to the next log";
goto err;
}
rli->event_relay_log_pos = BIN_LOG_HEADER_SIZE;
strmake(rli->event_relay_log_name,rli->linfo.log_file_name,
sizeof(rli->event_relay_log_name)-1);
flush_relay_log_info(rli);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
}
Fix for BUG#2011 "rare race condition producing "binlog has bad magic number" error in slave". The problem was that when the slave SQL thread reads a hot relay log (hot = the one being written to by the slave I/O thread), it must have the LOCK_log. It already took it for read_log_event(), but needs it also for check_binlog_magic(). This should fix all recently reported failures of the rpl_max_relay_size test in 4.1 and 5.0 (though the bug exists since 4.0, it showed up first in 5.0).
2003-12-04 15:30:14 +01:00
/*
Now we want to open this next log. To know if it's a hot log (the one
being written by the I/O thread now) or a cold log, we can use
is_active(); if it is hot, we use the I/O cache; if it's cold we open
the file normally. But if is_active() reports that the log is hot, this
may change between the test and the consequence of the test. So we may
open the I/O cache whereas the log is now cold, which is nonsense.
To guard against this, we need to have LOCK_log.
*/
DBUG_PRINT("info",("hot_log: %d",hot_log));
if (!hot_log) /* if hot_log, we already have this mutex */
pthread_mutex_lock(log_lock);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
if (rli->relay_log.is_active(rli->linfo.log_file_name))
{
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
#ifdef EXTRA_DEBUG
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
if (global_system_variables.log_warnings)
sql_print_information("next log '%s' is currently active",
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
rli->linfo.log_file_name);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
#endif
rli->cur_log= cur_log= rli->relay_log.get_log_file();
rli->cur_log_old_open_count= rli->relay_log.get_open_count();
DBUG_ASSERT(rli->cur_log_fd == -1);
/*
Read pointer has to be at the start since we are the only
reader.
Fix for BUG#2011 "rare race condition producing "binlog has bad magic number" error in slave". The problem was that when the slave SQL thread reads a hot relay log (hot = the one being written to by the slave I/O thread), it must have the LOCK_log. It already took it for read_log_event(), but needs it also for check_binlog_magic(). This should fix all recently reported failures of the rpl_max_relay_size test in 4.1 and 5.0 (though the bug exists since 4.0, it showed up first in 5.0).
2003-12-04 15:30:14 +01:00
We must keep the LOCK_log to read the 4 first bytes, as this is a hot
log (same as when we call read_log_event() above: for a hot log we
take the mutex).
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
*/
if (check_binlog_magic(cur_log,&errmsg))
Fix for BUG#2011 "rare race condition producing "binlog has bad magic number" error in slave". The problem was that when the slave SQL thread reads a hot relay log (hot = the one being written to by the slave I/O thread), it must have the LOCK_log. It already took it for read_log_event(), but needs it also for check_binlog_magic(). This should fix all recently reported failures of the rpl_max_relay_size test in 4.1 and 5.0 (though the bug exists since 4.0, it showed up first in 5.0).
2003-12-04 15:30:14 +01:00
{
if (!hot_log) pthread_mutex_unlock(log_lock);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
goto err;
Fix for BUG#2011 "rare race condition producing "binlog has bad magic number" error in slave". The problem was that when the slave SQL thread reads a hot relay log (hot = the one being written to by the slave I/O thread), it must have the LOCK_log. It already took it for read_log_event(), but needs it also for check_binlog_magic(). This should fix all recently reported failures of the rpl_max_relay_size test in 4.1 and 5.0 (though the bug exists since 4.0, it showed up first in 5.0).
2003-12-04 15:30:14 +01:00
}
if (!hot_log) pthread_mutex_unlock(log_lock);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
continue;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
Fix for BUG#2011 "rare race condition producing "binlog has bad magic number" error in slave". The problem was that when the slave SQL thread reads a hot relay log (hot = the one being written to by the slave I/O thread), it must have the LOCK_log. It already took it for read_log_event(), but needs it also for check_binlog_magic(). This should fix all recently reported failures of the rpl_max_relay_size test in 4.1 and 5.0 (though the bug exists since 4.0, it showed up first in 5.0).
2003-12-04 15:30:14 +01:00
if (!hot_log) pthread_mutex_unlock(log_lock);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
if we get here, the log was not hot, so we will have to open it
ourselves. We are sure that the log is still not hot now (a log can get
from hot to cold, but not from cold to hot). No need for LOCK_log.
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
*/
#ifdef EXTRA_DEBUG
Fixed many compiler warnings Fixed bugs in group_concat with ORDER BY and DISTINCT (Bugs #2695, #3381 and #3319) Fixed crash when doing rollback in slave and the io thread catched up with the sql thread Set locked_in_memory properly
2004-04-05 12:56:05 +02:00
if (global_system_variables.log_warnings)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
sql_print_information("next log '%s' is not active",
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
rli->linfo.log_file_name);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
#endif
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
// open_binlog() will check the magic header
if ((rli->cur_log_fd=open_binlog(cur_log,rli->linfo.log_file_name,
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
&errmsg)) <0)
goto err;
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
else
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
{
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
/*
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
Read failed with a non-EOF error.
TODO: come up with something better to handle this error
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
*/
if (hot_log)
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
pthread_mutex_unlock(log_lock);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
sql_print_error("Slave SQL thread: I/O error reading \
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
event(errno: %d cur_log->error: %d)",
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
my_errno,cur_log->error);
fixed replication bug not yet tested - will not be pushed for a few hours - need to commit in order to pull
2002-03-09 21:50:07 +01:00
// set read position to the beginning of the event
Replication: new code to not modify in-memory log positions until the COMMIT is executed, even if the transaction spans on >=2 relay logs (bug #53). New variable relay_log_purge =0|1 New test to verify bug #53
2003-04-24 15:29:25 +02:00
my_b_seek(cur_log,rli->event_relay_log_pos);
reckless slave option bison 1.34 yacc rule compliance more robust handling of relay log corruption
2002-04-16 01:09:30 +02:00
/* otherwise, we have had a partial read */
errmsg = "Aborting slave SQL thread because of partial event read";
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
break; // To end of function
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
}
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
if (!errmsg && global_system_variables.log_warnings)
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
{
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
sql_print_information("Error reading relay log event: %s",
Added options --auto-increment-increment and --auto-increment-offset. This allows one to setup a master <-> master replication with non conflicting auto-increment series. Cleaned up binary log code to make it easyer to add new state variables. Added simpler 'upper level' logic for artificial events (events that should not cause cleanups on slave). Simplified binary log handling. Changed how auto_increment works together with to SET INSERT_ID=# to make it more predictable: Now the inserted rows in a multi-row statement are set independent of the existing rows in the table. (Before only InnoDB did this correctly)
2004-09-15 21:10:31 +02:00
"slave SQL thread was killed");
DBUG_RETURN(0);
}
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
err:
Lots of code fixes to the replication code (especially the binary logging and index log file handling) Fixed bugs in my last changeset that made MySQL hard to compile. Added mutex around some data that could cause table cache corruptions when using OPTIMIZE TABLE / REPAIR TABLE or automatic repair of MyISAM tables. Added mutex around some data in the slave start/stop code that could cause THD linked list corruptions Extended my_chsize() to allow one to specify a filler character. Extend vio_blocking to return the old state (This made some usage of this function much simpler) Added testing for some functions that they caller have got the required mutexes before calling the function. Use setrlimit() to ensure that we can write core file if one specifies --core-file. Added --slave-compressed-protocol Made 2 the minimum length for ft_min_word_len Added variables foreign_key_checks & unique_checks. Less logging from replication code (if not started with --log-warnings) Changed that SHOW INNODB STATUS requre the SUPER privilege More DBUG statements and a lot of new code comments
2002-08-08 02:12:02 +02:00
if (errmsg)
sql_print_error("Error reading relay log event: %s", errmsg);
removed init_count from IO_CACHE. Added missing mutex_unlock to slave replication code.
2002-06-05 22:04:38 +02:00
DBUG_RETURN(0);
Here comes a nasty patch, although I am not ready to push it yet. I will first pull, merge,test, and get it to work. The main change is the new replication code - now we have two slave threads SQL thread and I/O thread. I have also re-written a lot of the code to prepare for multi-master implementation. I also documented IO_CACHE quite extensively and to some extend, THD class.
2002-01-20 03:16:52 +01:00
}
WL#912 (more user control on relay logs): FLUSH LOGS now rotates relay logs, and a new variable max_relay_log_size. Plus a very small bit of code cleaning.
2003-07-06 17:59:54 +02:00
/*
Rotate a relay log (this is used only by FLUSH LOGS; the automatic rotation
because of size is simpler because when we do it we already have all relevant
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
locks; here we don't, so this function is mainly taking locks).
WL#3153 "Split logs". Recommit with post-review fixes
2006-05-05 08:45:58 +02:00
Returns nothing as we cannot catch any error (MYSQL_BIN_LOG::new_file()
is void).
WL#912 (more user control on relay logs): FLUSH LOGS now rotates relay logs, and a new variable max_relay_log_size. Plus a very small bit of code cleaning.
2003-07-06 17:59:54 +02:00
*/
Renaming MASTER_INFO to Master_info in order to follow the coding standards (and help Doxygen generating good documentation).
2007-08-16 08:52:50 +02:00
void rotate_relay_log(Master_info* mi)
WL#912 (more user control on relay logs): FLUSH LOGS now rotates relay logs, and a new variable max_relay_log_size. Plus a very small bit of code cleaning.
2003-07-06 17:59:54 +02:00
{
DBUG_ENTER("rotate_relay_log");
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
Relay_log_info* rli= &mi->rli;
Fix for BUG#791: a safer way of initing the mutexes in MYSQL_LOG. is_open() is now always thread-safe. See each file for details.
2003-07-11 14:26:44 +02:00
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
/* We don't lock rli->run_lock. This would lead to deadlocks. */
pthread_mutex_lock(&mi->run_lock);
slave.cc: Tidy up white space.
2006-07-07 09:27:55 +02:00
/*
Fix for BUG#791: a safer way of initing the mutexes in MYSQL_LOG. is_open() is now always thread-safe. See each file for details.
2003-07-11 14:26:44 +02:00
We need to test inited because otherwise, new_file() will attempt to lock
LOCK_log, which may not be inited (if we're not a slave).
*/
WL#912 (more user control on relay logs): FLUSH LOGS now rotates relay logs, and a new variable max_relay_log_size. Plus a very small bit of code cleaning.
2003-07-06 17:59:54 +02:00
if (!rli->inited)
{
Cleanups
2003-07-08 23:55:07 +02:00
DBUG_PRINT("info", ("rli->inited == 0"));
Fix for BUG#791: a safer way of initing the mutexes in MYSQL_LOG. is_open() is now always thread-safe. See each file for details.
2003-07-11 14:26:44 +02:00
goto end;
WL#912 (more user control on relay logs): FLUSH LOGS now rotates relay logs, and a new variable max_relay_log_size. Plus a very small bit of code cleaning.
2003-07-06 17:59:54 +02:00
}
Cleanups
2003-07-08 23:55:07 +02:00
WL#912 (more user control on relay logs): FLUSH LOGS now rotates relay logs, and a new variable max_relay_log_size. Plus a very small bit of code cleaning.
2003-07-06 17:59:54 +02:00
/* If the relay log is closed, new_file() will do nothing. */
WL#3153 "Split logs". Recommit with post-review fixes
2006-05-05 08:45:58 +02:00
rli->relay_log.new_file();
Cleanups
2003-07-08 23:55:07 +02:00
WL#912 (more user control on relay logs): FLUSH LOGS now rotates relay logs, and a new variable max_relay_log_size. Plus a very small bit of code cleaning.
2003-07-06 17:59:54 +02:00
/*
We harvest now, because otherwise BIN_LOG_HEADER_SIZE will not immediately
be counted, so imagine a succession of FLUSH LOGS and assume the slave
threads are started:
Cleanups
2003-07-08 23:55:07 +02:00
relay_log_space decreases by the size of the deleted relay log, but does
not increase, so flush-after-flush we may become negative, which is wrong.
Even if this will be corrected as soon as a query is replicated on the
slave (because the I/O thread will then call harvest_bytes_written() which
will harvest all these BIN_LOG_HEADER_SIZE we forgot), it may give strange
output in SHOW SLAVE STATUS meanwhile. So we harvest now.
WL#912 (more user control on relay logs): FLUSH LOGS now rotates relay logs, and a new variable max_relay_log_size. Plus a very small bit of code cleaning.
2003-07-06 17:59:54 +02:00
If the log is closed, then this will just harvest the last writes, probably
0 as they probably have been harvested.
*/
rli->relay_log.harvest_bytes_written(&rli->log_space_total);
Fix for BUG#791: a safer way of initing the mutexes in MYSQL_LOG. is_open() is now always thread-safe. See each file for details.
2003-07-11 14:26:44 +02:00
end:
This will be pushed only after I fix the testsuite. This is the main commit for Worklog tasks: * A more dynamic binlog format which allows small changes (1064) * Log session variables in Query_log_event (1063) Below 5.0 means 5.0.0. MySQL 5.0 is able to replicate FOREIGN_KEY_CHECKS, UNIQUE_KEY_CHECKS (for speed), SQL_AUTO_IS_NULL, SQL_MODE. Not charsets (WL#1062), not some vars (I can only think of SQL_SELECT_LIMIT, which deserves a special treatment). Note that this works for queries, except LOAD DATA INFILE (for this it would have to wait for Dmitri's push of WL#874, which in turns waits for the present push, so... the deadlock must be broken!). Note that when Dmitri pushes WL#874 in 5.0.1, 5.0.0 won't be able to replicate a LOAD DATA INFILE from 5.0.1. Apart from that, the new binlog format is designed so that it can tolerate a little variation in the events (so that a 5.0.0 slave could replicate a 5.0.1 master, except for LOAD DATA INFILE unfortunately); that is, when I later add replication of charsets it should break nothing. And when I later add a UID to every event, it should break nothing. The main change brought by this patch is a new type of event, Format_description_log_event, which describes some lengthes in other event types. This event is needed for the master/slave/mysqlbinlog to understand a 5.0 log. Thanks to this event, we can later add more bytes to the header of every event without breaking compatibility. Inside Query_log_event, we have some additional dynamic format, as every Query_log_event can have a different number of status variables, stored as pairs (code, value); that's how SQL_MODE and session variables and catalog are stored. Like this, we can later add count of affected rows, charsets... and we can have options --don't-log-count-affected-rows if we want. MySQL 5.0 is able to run on 4.x relay logs, 4.x binlogs. Upgrading a 4.x master to 5.0 is ok (no need to delete binlogs), upgrading a 4.x slave to 5.0 is ok (no need to delete relay logs); so both can be "hot" upgrades. Upgrading a 3.23 master to 5.0 requires as much as upgrading it to 4.0. 3.23 and 4.x can't be slaves of 5.0. So downgrading from 5.0 to 4.x may be complicated. Log_event::log_pos is now the position of the end of the event, which is more useful than the position of the beginning. We take care about compatibility with <5.0 (in which log_pos is the beginning). I added a short test for replication of SQL_MODE and some other variables. TODO: - after committing this, merge the latest 5.0 into it - fix all tests - update the manual with upgrade notes.
2003-12-18 01:09:05 +01:00
pthread_mutex_unlock(&mi->run_lock);
WL#912 (more user control on relay logs): FLUSH LOGS now rotates relay logs, and a new variable max_relay_log_size. Plus a very small bit of code cleaning.
2003-07-06 17:59:54 +02:00
DBUG_VOID_RETURN;
}
cleanup removal of duplicate code in mf_iocache.cc work on failsafe replication work on SEQ_READ_APPEND io cache
2001-10-23 21:28:03 +02:00
Fix for BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values". When in an INSERT ON DUPLICATE KEY UPDATE, using an autoincrement column, we inserted some autogenerated values and also updated some rows, some autogenerated values were not used (for example, even if 10 was the largest autoinc value in the table at the start of the statement, 12 could be the first autogenerated value inserted by the statement, instead of 11). One autogenerated value was lost per updated row. Led to exhausting the range of the autoincrement column faster. Bug introduced by fix of BUG#20188; present since 5.0.24 and 5.1.12. This bug breaks replication from a pre-5.0.24 master. But the present bugfix, as it makes INSERT ON DUP KEY UPDATE behave like pre-5.0.24, breaks replication from a [5.0.24,5.0.34] master to a fixed (5.0.36) slave! To warn users against this when they upgrade their slave, as agreed with the support team, we add code for a fixed slave to detect that it is connected to a buggy master in a situation (INSERT ON DUP KEY UPDATE into autoinc column) likely to break replication, in which case it cannot replicate so stops and prints a message to the slave's error log and to SHOW SLAVE STATUS. For 5.0.36->[5.0.24,5.0.34] replication we cannot warn as master does not know the slave's version (but we always recommended to users to have slave at least as new as master). As agreed with support, I'll also ask for an alert to be put into the MySQL Network Monitoring and Advisory Service.
2007-02-08 15:53:14 +01:00
/**
Detects, based on master's version (as found in the relay log), if master
has a certain bug.
Renaming RELAY_LOG_INFO and st_relay_log_info to follow coding standards (and be more friendly to Doxygen by removing unnecessary typedefs).
2007-08-16 07:37:50 +02:00
@param rli Relay_log_info which tells the master's version
Fix for BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values". When in an INSERT ON DUPLICATE KEY UPDATE, using an autoincrement column, we inserted some autogenerated values and also updated some rows, some autogenerated values were not used (for example, even if 10 was the largest autoinc value in the table at the start of the statement, 12 could be the first autogenerated value inserted by the statement, instead of 11). One autogenerated value was lost per updated row. Led to exhausting the range of the autoincrement column faster. Bug introduced by fix of BUG#20188; present since 5.0.24 and 5.1.12. This bug breaks replication from a pre-5.0.24 master. But the present bugfix, as it makes INSERT ON DUP KEY UPDATE behave like pre-5.0.24, breaks replication from a [5.0.24,5.0.34] master to a fixed (5.0.36) slave! To warn users against this when they upgrade their slave, as agreed with the support team, we add code for a fixed slave to detect that it is connected to a buggy master in a situation (INSERT ON DUP KEY UPDATE into autoinc column) likely to break replication, in which case it cannot replicate so stops and prints a message to the slave's error log and to SHOW SLAVE STATUS. For 5.0.36->[5.0.24,5.0.34] replication we cannot warn as master does not know the slave's version (but we always recommended to users to have slave at least as new as master). As agreed with support, I'll also ask for an alert to be put into the MySQL Network Monitoring and Advisory Service.
2007-02-08 15:53:14 +01:00
@param bug_id Number of the bug as found in bugs.mysql.com
BUG#33029 5.0 to 5.1 replication fails on dup key when inserting using a trig in SP For all 5.0 and up to 5.1.12 exclusive, when a stored routine or trigger caused an INSERT into an AUTO_INCREMENT column, the generated AUTO_INCREMENT value should not be written into the binary log, which means if a statement does not generate AUTO_INCREMENT value itself, there will be no Intvar event (SET INSERT_ID) associated with it even if one of the stored routine or trigger caused generation of such a value. And meanwhile, when executing a stored routine or trigger, it would ignore the INSERT_ID value even if there is a INSERT_ID value available set by a SET INSERT_ID statement. Starting from MySQL 5.1.12, the generated AUTO_INCREMENT value is written into the binary log, and the value will be used if available when executing the stored routine or trigger. Prior fix of this bug in MySQL 5.0 and prior MySQL 5.1.12 (referenced as the buggy versions in the text below), when a statement that generates AUTO_INCREMENT value by the top statement was executed in the body of a SP, all statements in the SP after this statement would be treated as if they had generated AUTO_INCREMENT by the top statement. When a statement that did not generate AUTO_INCREMENT value by the top statement but by a function/trigger called by it, an erroneous Intvar event would be associated with the statement, this erroneous INSERT_ID value wouldn't cause problem when replicating between masters and slaves of 5.0.x or prior 5.1.12, because the erroneous INSERT_ID value was not used when executing functions/triggers. But when replicating from buggy versions to 5.1.12 or newer, which will use the INSERT_ID value in functions/triggers, the erroneous value will be used, which would cause duplicate entry error and cause the slave to stop. The patch for 5.1 fixed it to ignore the SET INSERT_ID value when executing functions/triggers if it is replicating from a master of buggy versions, another patch for 5.0 fixed it not to generate the erroneous Intvar event.
2008-03-14 04:35:41 +01:00
@param report bool report error message, default TRUE
BUG#37426: RBR breaks for CHAR() UTF-8 fields > 85 chars In order to handle CHAR() fields, 8 bits were reserved for the size of the CHAR field. However, instead of denoting the number of characters in the field, field_length was used which denotes the number of bytes in the field. Since UTF-8 fields can have three bytes per character (and has been extended to have four bytes per character in 6.0), an extra two bits have been encoded in the field metadata work for fields of type Field_string (i.e., CHAR fields). Since the metadata word is filled, the extra bits have been encoded in the upper 4 bits of the real type (the most significant byte of the metadata word) by computing the bitwise xor of the extra two bits. Since the upper 4 bits of the real type always is 1111 for Field_string, this means that for fields of length <256, the encoding is identical to the encoding used in pre-5.1.26 servers, but for lengths of 256 or more, an unrecognized type is formed, causing an old slave (that does not handle lengths of 256 or more) to stop.
2008-06-30 22:11:18 +02:00
@param pred Predicate function that will be called with @c param to
check for the bug. If the function return @c true, the bug is present,
otherwise, it is not.
@param param State passed to @c pred function.
Fix for BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values". When in an INSERT ON DUPLICATE KEY UPDATE, using an autoincrement column, we inserted some autogenerated values and also updated some rows, some autogenerated values were not used (for example, even if 10 was the largest autoinc value in the table at the start of the statement, 12 could be the first autogenerated value inserted by the statement, instead of 11). One autogenerated value was lost per updated row. Led to exhausting the range of the autoincrement column faster. Bug introduced by fix of BUG#20188; present since 5.0.24 and 5.1.12. This bug breaks replication from a pre-5.0.24 master. But the present bugfix, as it makes INSERT ON DUP KEY UPDATE behave like pre-5.0.24, breaks replication from a [5.0.24,5.0.34] master to a fixed (5.0.36) slave! To warn users against this when they upgrade their slave, as agreed with the support team, we add code for a fixed slave to detect that it is connected to a buggy master in a situation (INSERT ON DUP KEY UPDATE into autoinc column) likely to break replication, in which case it cannot replicate so stops and prints a message to the slave's error log and to SHOW SLAVE STATUS. For 5.0.36->[5.0.24,5.0.34] replication we cannot warn as master does not know the slave's version (but we always recommended to users to have slave at least as new as master). As agreed with support, I'll also ask for an alert to be put into the MySQL Network Monitoring and Advisory Service.
2007-02-08 15:53:14 +01:00
@return TRUE if master has the bug, FALSE if it does not.
*/
BUG#37426: RBR breaks for CHAR() UTF-8 fields > 85 chars In order to handle CHAR() fields, 8 bits were reserved for the size of the CHAR field. However, instead of denoting the number of characters in the field, field_length was used which denotes the number of bytes in the field. Since UTF-8 fields can have three bytes per character (and has been extended to have four bytes per character in 6.0), an extra two bits have been encoded in the field metadata work for fields of type Field_string (i.e., CHAR fields). Since the metadata word is filled, the extra bits have been encoded in the upper 4 bits of the real type (the most significant byte of the metadata word) by computing the bitwise xor of the extra two bits. Since the upper 4 bits of the real type always is 1111 for Field_string, this means that for fields of length <256, the encoding is identical to the encoding used in pre-5.1.26 servers, but for lengths of 256 or more, an unrecognized type is formed, causing an old slave (that does not handle lengths of 256 or more) to stop.
2008-06-30 22:11:18 +02:00
bool rpl_master_has_bug(const Relay_log_info *rli, uint bug_id, bool report,
bool (*pred)(const void *), const void *param)
Fix for BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values". When in an INSERT ON DUPLICATE KEY UPDATE, using an autoincrement column, we inserted some autogenerated values and also updated some rows, some autogenerated values were not used (for example, even if 10 was the largest autoinc value in the table at the start of the statement, 12 could be the first autogenerated value inserted by the statement, instead of 11). One autogenerated value was lost per updated row. Led to exhausting the range of the autoincrement column faster. Bug introduced by fix of BUG#20188; present since 5.0.24 and 5.1.12. This bug breaks replication from a pre-5.0.24 master. But the present bugfix, as it makes INSERT ON DUP KEY UPDATE behave like pre-5.0.24, breaks replication from a [5.0.24,5.0.34] master to a fixed (5.0.36) slave! To warn users against this when they upgrade their slave, as agreed with the support team, we add code for a fixed slave to detect that it is connected to a buggy master in a situation (INSERT ON DUP KEY UPDATE into autoinc column) likely to break replication, in which case it cannot replicate so stops and prints a message to the slave's error log and to SHOW SLAVE STATUS. For 5.0.36->[5.0.24,5.0.34] replication we cannot warn as master does not know the slave's version (but we always recommended to users to have slave at least as new as master). As agreed with support, I'll also ask for an alert to be put into the MySQL Network Monitoring and Advisory Service.
2007-02-08 15:53:14 +01:00
{
struct st_version_range_for_one_bug {
uint bug_id;
const uchar introduced_in[3]; // first version with bug
const uchar fixed_in[3]; // first version with fix
};
static struct st_version_range_for_one_bug versions_for_all_bugs[]=
{
the fix for BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values" didn't make it into 5.0.36 and 5.1.16, so we need to adjust the bug-detection-based-on-version-number code. Because the rpl tree has a too old version, rpl_insert_id cannot pass, so I disable it (like is already the case in 5.1-rpl for the same reason), and the repl team will re-enable it when they merge 5.0 and 5.1 into their trees (thus getting the right version number).
2007-02-23 15:32:51 +01:00
{24432, { 5, 0, 24 }, { 5, 0, 38 } },
BUG#33029 5.0 to 5.1 replication fails on dup key when inserting using a trig in SP For all 5.0 and up to 5.1.12 exclusive, when a stored routine or trigger caused an INSERT into an AUTO_INCREMENT column, the generated AUTO_INCREMENT value should not be written into the binary log, which means if a statement does not generate AUTO_INCREMENT value itself, there will be no Intvar event (SET INSERT_ID) associated with it even if one of the stored routine or trigger caused generation of such a value. And meanwhile, when executing a stored routine or trigger, it would ignore the INSERT_ID value even if there is a INSERT_ID value available set by a SET INSERT_ID statement. Starting from MySQL 5.1.12, the generated AUTO_INCREMENT value is written into the binary log, and the value will be used if available when executing the stored routine or trigger. Prior fix of this bug in MySQL 5.0 and prior MySQL 5.1.12 (referenced as the buggy versions in the text below), when a statement that generates AUTO_INCREMENT value by the top statement was executed in the body of a SP, all statements in the SP after this statement would be treated as if they had generated AUTO_INCREMENT by the top statement. When a statement that did not generate AUTO_INCREMENT value by the top statement but by a function/trigger called by it, an erroneous Intvar event would be associated with the statement, this erroneous INSERT_ID value wouldn't cause problem when replicating between masters and slaves of 5.0.x or prior 5.1.12, because the erroneous INSERT_ID value was not used when executing functions/triggers. But when replicating from buggy versions to 5.1.12 or newer, which will use the INSERT_ID value in functions/triggers, the erroneous value will be used, which would cause duplicate entry error and cause the slave to stop. The patch for 5.1 fixed it to ignore the SET INSERT_ID value when executing functions/triggers if it is replicating from a master of buggy versions, another patch for 5.0 fixed it not to generate the erroneous Intvar event.
2008-03-14 04:35:41 +01:00
{24432, { 5, 1, 12 }, { 5, 1, 17 } },
{33029, { 5, 0, 0 }, { 5, 0, 58 } },
{33029, { 5, 1, 0 }, { 5, 1, 12 } },
BUG#37426: RBR breaks for CHAR() UTF-8 fields > 85 chars In order to handle CHAR() fields, 8 bits were reserved for the size of the CHAR field. However, instead of denoting the number of characters in the field, field_length was used which denotes the number of bytes in the field. Since UTF-8 fields can have three bytes per character (and has been extended to have four bytes per character in 6.0), an extra two bits have been encoded in the field metadata work for fields of type Field_string (i.e., CHAR fields). Since the metadata word is filled, the extra bits have been encoded in the upper 4 bits of the real type (the most significant byte of the metadata word) by computing the bitwise xor of the extra two bits. Since the upper 4 bits of the real type always is 1111 for Field_string, this means that for fields of length <256, the encoding is identical to the encoding used in pre-5.1.26 servers, but for lengths of 256 or more, an unrecognized type is formed, causing an old slave (that does not handle lengths of 256 or more) to stop.
2008-06-30 22:11:18 +02:00
{37426, { 5, 1, 0 }, { 5, 1, 26 } },
Fix for BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values". When in an INSERT ON DUPLICATE KEY UPDATE, using an autoincrement column, we inserted some autogenerated values and also updated some rows, some autogenerated values were not used (for example, even if 10 was the largest autoinc value in the table at the start of the statement, 12 could be the first autogenerated value inserted by the statement, instead of 11). One autogenerated value was lost per updated row. Led to exhausting the range of the autoincrement column faster. Bug introduced by fix of BUG#20188; present since 5.0.24 and 5.1.12. This bug breaks replication from a pre-5.0.24 master. But the present bugfix, as it makes INSERT ON DUP KEY UPDATE behave like pre-5.0.24, breaks replication from a [5.0.24,5.0.34] master to a fixed (5.0.36) slave! To warn users against this when they upgrade their slave, as agreed with the support team, we add code for a fixed slave to detect that it is connected to a buggy master in a situation (INSERT ON DUP KEY UPDATE into autoinc column) likely to break replication, in which case it cannot replicate so stops and prints a message to the slave's error log and to SHOW SLAVE STATUS. For 5.0.36->[5.0.24,5.0.34] replication we cannot warn as master does not know the slave's version (but we always recommended to users to have slave at least as new as master). As agreed with support, I'll also ask for an alert to be put into the MySQL Network Monitoring and Advisory Service.
2007-02-08 15:53:14 +01:00
};
const uchar *master_ver=
rli->relay_log.description_event_for_exec->server_version_split;
DBUG_ASSERT(sizeof(rli->relay_log.description_event_for_exec->server_version_split) == 3);
for (uint i= 0;
i < sizeof(versions_for_all_bugs)/sizeof(*versions_for_all_bugs);i++)
{
const uchar *introduced_in= versions_for_all_bugs[i].introduced_in,
*fixed_in= versions_for_all_bugs[i].fixed_in;
if ((versions_for_all_bugs[i].bug_id == bug_id) &&
(memcmp(introduced_in, master_ver, 3) <= 0) &&
BUG#37426: RBR breaks for CHAR() UTF-8 fields > 85 chars In order to handle CHAR() fields, 8 bits were reserved for the size of the CHAR field. However, instead of denoting the number of characters in the field, field_length was used which denotes the number of bytes in the field. Since UTF-8 fields can have three bytes per character (and has been extended to have four bytes per character in 6.0), an extra two bits have been encoded in the field metadata work for fields of type Field_string (i.e., CHAR fields). Since the metadata word is filled, the extra bits have been encoded in the upper 4 bits of the real type (the most significant byte of the metadata word) by computing the bitwise xor of the extra two bits. Since the upper 4 bits of the real type always is 1111 for Field_string, this means that for fields of length <256, the encoding is identical to the encoding used in pre-5.1.26 servers, but for lengths of 256 or more, an unrecognized type is formed, causing an old slave (that does not handle lengths of 256 or more) to stop.
2008-06-30 22:11:18 +02:00
(memcmp(fixed_in, master_ver, 3) > 0) &&
(pred == NULL || (*pred)(param)))
Fix for BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values". When in an INSERT ON DUPLICATE KEY UPDATE, using an autoincrement column, we inserted some autogenerated values and also updated some rows, some autogenerated values were not used (for example, even if 10 was the largest autoinc value in the table at the start of the statement, 12 could be the first autogenerated value inserted by the statement, instead of 11). One autogenerated value was lost per updated row. Led to exhausting the range of the autoincrement column faster. Bug introduced by fix of BUG#20188; present since 5.0.24 and 5.1.12. This bug breaks replication from a pre-5.0.24 master. But the present bugfix, as it makes INSERT ON DUP KEY UPDATE behave like pre-5.0.24, breaks replication from a [5.0.24,5.0.34] master to a fixed (5.0.36) slave! To warn users against this when they upgrade their slave, as agreed with the support team, we add code for a fixed slave to detect that it is connected to a buggy master in a situation (INSERT ON DUP KEY UPDATE into autoinc column) likely to break replication, in which case it cannot replicate so stops and prints a message to the slave's error log and to SHOW SLAVE STATUS. For 5.0.36->[5.0.24,5.0.34] replication we cannot warn as master does not know the slave's version (but we always recommended to users to have slave at least as new as master). As agreed with support, I'll also ask for an alert to be put into the MySQL Network Monitoring and Advisory Service.
2007-02-08 15:53:14 +01:00
{
BUG#33029 5.0 to 5.1 replication fails on dup key when inserting using a trig in SP For all 5.0 and up to 5.1.12 exclusive, when a stored routine or trigger caused an INSERT into an AUTO_INCREMENT column, the generated AUTO_INCREMENT value should not be written into the binary log, which means if a statement does not generate AUTO_INCREMENT value itself, there will be no Intvar event (SET INSERT_ID) associated with it even if one of the stored routine or trigger caused generation of such a value. And meanwhile, when executing a stored routine or trigger, it would ignore the INSERT_ID value even if there is a INSERT_ID value available set by a SET INSERT_ID statement. Starting from MySQL 5.1.12, the generated AUTO_INCREMENT value is written into the binary log, and the value will be used if available when executing the stored routine or trigger. Prior fix of this bug in MySQL 5.0 and prior MySQL 5.1.12 (referenced as the buggy versions in the text below), when a statement that generates AUTO_INCREMENT value by the top statement was executed in the body of a SP, all statements in the SP after this statement would be treated as if they had generated AUTO_INCREMENT by the top statement. When a statement that did not generate AUTO_INCREMENT value by the top statement but by a function/trigger called by it, an erroneous Intvar event would be associated with the statement, this erroneous INSERT_ID value wouldn't cause problem when replicating between masters and slaves of 5.0.x or prior 5.1.12, because the erroneous INSERT_ID value was not used when executing functions/triggers. But when replicating from buggy versions to 5.1.12 or newer, which will use the INSERT_ID value in functions/triggers, the erroneous value will be used, which would cause duplicate entry error and cause the slave to stop. The patch for 5.1 fixed it to ignore the SET INSERT_ID value when executing functions/triggers if it is replicating from a master of buggy versions, another patch for 5.0 fixed it not to generate the erroneous Intvar event.
2008-03-14 04:35:41 +01:00
if (!report)
return TRUE;
Fix for BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values". When in an INSERT ON DUPLICATE KEY UPDATE, using an autoincrement column, we inserted some autogenerated values and also updated some rows, some autogenerated values were not used (for example, even if 10 was the largest autoinc value in the table at the start of the statement, 12 could be the first autogenerated value inserted by the statement, instead of 11). One autogenerated value was lost per updated row. Led to exhausting the range of the autoincrement column faster. Bug introduced by fix of BUG#20188; present since 5.0.24 and 5.1.12. This bug breaks replication from a pre-5.0.24 master. But the present bugfix, as it makes INSERT ON DUP KEY UPDATE behave like pre-5.0.24, breaks replication from a [5.0.24,5.0.34] master to a fixed (5.0.36) slave! To warn users against this when they upgrade their slave, as agreed with the support team, we add code for a fixed slave to detect that it is connected to a buggy master in a situation (INSERT ON DUP KEY UPDATE into autoinc column) likely to break replication, in which case it cannot replicate so stops and prints a message to the slave's error log and to SHOW SLAVE STATUS. For 5.0.36->[5.0.24,5.0.34] replication we cannot warn as master does not know the slave's version (but we always recommended to users to have slave at least as new as master). As agreed with support, I'll also ask for an alert to be put into the MySQL Network Monitoring and Advisory Service.
2007-02-08 15:53:14 +01:00
// a short message for SHOW SLAVE STATUS (message length constraints)
my_printf_error(ER_UNKNOWN_ERROR, "master may suffer from"
" http://bugs.mysql.com/bug.php?id=%u"
" so slave stops; check error log on slave"
" for more info", MYF(0), bug_id);
Manual merge from 5.0-rpl, of fixes for: 1) BUG#25507 "multi-row insert delayed + auto increment causes duplicate key entries on slave" (two concurrrent connections doing multi-row INSERT DELAYED to insert into an auto_increment column, caused replication slave to stop with "duplicate key error" (and binlog was wrong), and BUG#26116 "If multi-row INSERT DELAYED has errors, statement-based binlogging breaks" (the binlog was not accounting for all rows inserted, or slave could stop). The fix is that: in statement-based binlogging, a multi-row INSERT DELAYED is silently converted to a non-delayed INSERT. This is supposed to not affect many 5.1 users as in 5.1, the default binlog format is "mixed", which does not have the bug (the bug is only with binlog_format=STATEMENT). We should document how the system delayed_insert thread decides of its binlog format (which is not modified by this patch): this decision is taken when the thread is created and holds until it is terminated (is not affected by any later change via SET GLOBAL BINLOG_FORMAT). It is also not affected by the binlog format of the connection which issues INSERT DELAYED (this binlog format does not affect how the row will be binlogged). If one wants to change the binlog format of its server with SET GLOBAL BINLOG_FORMAT, it should do FLUSH TABLES to be sure all delayed_insert threads terminate and thus new threads are created, taking into account the new format. 2) BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values". When in an INSERT ON DUPLICATE KEY UPDATE, using an autoincrement column, we inserted some autogenerated values and also updated some rows, some autogenerated values were not used (for example, even if 10 was the largest autoinc value in the table at the start of the statement, 12 could be the first autogenerated value inserted by the statement, instead of 11). One autogenerated value was lost per updated row. Led to exhausting the range of the autoincrement column faster. Bug introduced by fix of BUG#20188; present since 5.0.24 and 5.1.12. This bug breaks replication from a pre-5.0.24/pre-5.1.12 master. But the present bugfix, as it makes INSERT ON DUP KEY UPDATE behave like pre-5.0.24/pre-5.1.12, breaks replication from a [5.0.24,5.0.34]/[5.1.12,5.1.15] master to a fixed (5.0.36/5.1.16) slave! To warn users against this when they upgrade their slave, as agreed with the support team, we add code for a fixed slave to detect that it is connected to a buggy master in a situation (INSERT ON DUP KEY UPDATE into autoinc column) likely to break replication, in which case it cannot replicate so stops and prints a message to the slave's error log and to SHOW SLAVE STATUS. For 5.0.36->[5.0.24,5.0.34] replication or 5.1.16->[5.1.12,5.1.15] replication we cannot warn as master does not know the slave's version (but we always recommended to users to have slave at least as new as master). As agreed with support, I have asked for an alert to be put into the MySQL Network Monitoring and Advisory Service. 3) note that I'll re-enable rpl_insert_id as soon as 5.1-rpl gets the changes from the main 5.1.
2007-02-15 20:28:58 +01:00
// a verbose message for the error log
BUG#24954 (Last_errno and Last_error not set after master_retry_count has been reached): Post-merge patch to handle all the changes to the tree since the tree was cloned.
2007-06-11 22:15:39 +02:00
rli->report(ERROR_LEVEL, ER_UNKNOWN_ERROR,
"According to the master's version ('%s'),"
" it is probable that master suffers from this bug:"
Manual merge from 5.0-rpl, of fixes for: 1) BUG#25507 "multi-row insert delayed + auto increment causes duplicate key entries on slave" (two concurrrent connections doing multi-row INSERT DELAYED to insert into an auto_increment column, caused replication slave to stop with "duplicate key error" (and binlog was wrong), and BUG#26116 "If multi-row INSERT DELAYED has errors, statement-based binlogging breaks" (the binlog was not accounting for all rows inserted, or slave could stop). The fix is that: in statement-based binlogging, a multi-row INSERT DELAYED is silently converted to a non-delayed INSERT. This is supposed to not affect many 5.1 users as in 5.1, the default binlog format is "mixed", which does not have the bug (the bug is only with binlog_format=STATEMENT). We should document how the system delayed_insert thread decides of its binlog format (which is not modified by this patch): this decision is taken when the thread is created and holds until it is terminated (is not affected by any later change via SET GLOBAL BINLOG_FORMAT). It is also not affected by the binlog format of the connection which issues INSERT DELAYED (this binlog format does not affect how the row will be binlogged). If one wants to change the binlog format of its server with SET GLOBAL BINLOG_FORMAT, it should do FLUSH TABLES to be sure all delayed_insert threads terminate and thus new threads are created, taking into account the new format. 2) BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values". When in an INSERT ON DUPLICATE KEY UPDATE, using an autoincrement column, we inserted some autogenerated values and also updated some rows, some autogenerated values were not used (for example, even if 10 was the largest autoinc value in the table at the start of the statement, 12 could be the first autogenerated value inserted by the statement, instead of 11). One autogenerated value was lost per updated row. Led to exhausting the range of the autoincrement column faster. Bug introduced by fix of BUG#20188; present since 5.0.24 and 5.1.12. This bug breaks replication from a pre-5.0.24/pre-5.1.12 master. But the present bugfix, as it makes INSERT ON DUP KEY UPDATE behave like pre-5.0.24/pre-5.1.12, breaks replication from a [5.0.24,5.0.34]/[5.1.12,5.1.15] master to a fixed (5.0.36/5.1.16) slave! To warn users against this when they upgrade their slave, as agreed with the support team, we add code for a fixed slave to detect that it is connected to a buggy master in a situation (INSERT ON DUP KEY UPDATE into autoinc column) likely to break replication, in which case it cannot replicate so stops and prints a message to the slave's error log and to SHOW SLAVE STATUS. For 5.0.36->[5.0.24,5.0.34] replication or 5.1.16->[5.1.12,5.1.15] replication we cannot warn as master does not know the slave's version (but we always recommended to users to have slave at least as new as master). As agreed with support, I have asked for an alert to be put into the MySQL Network Monitoring and Advisory Service. 3) note that I'll re-enable rpl_insert_id as soon as 5.1-rpl gets the changes from the main 5.1.
2007-02-15 20:28:58 +01:00
" http://bugs.mysql.com/bug.php?id=%u"
" and thus replicating the current binary log event"
" may make the slave's data become different from the"
" master's data."
" To take no risk, slave refuses to replicate"
" this event and stops."
" We recommend that all updates be stopped on the"
" master and slave, that the data of both be"
" manually synchronized,"
" that master's binary logs be deleted,"
" that master be upgraded to a version at least"
" equal to '%d.%d.%d'. Then replication can be"
" restarted.",
rli->relay_log.description_event_for_exec->server_version,
bug_id,
fixed_in[0], fixed_in[1], fixed_in[2]);
Fix for BUG#24432 "INSERT... ON DUPLICATE KEY UPDATE skips auto_increment values". When in an INSERT ON DUPLICATE KEY UPDATE, using an autoincrement column, we inserted some autogenerated values and also updated some rows, some autogenerated values were not used (for example, even if 10 was the largest autoinc value in the table at the start of the statement, 12 could be the first autogenerated value inserted by the statement, instead of 11). One autogenerated value was lost per updated row. Led to exhausting the range of the autoincrement column faster. Bug introduced by fix of BUG#20188; present since 5.0.24 and 5.1.12. This bug breaks replication from a pre-5.0.24 master. But the present bugfix, as it makes INSERT ON DUP KEY UPDATE behave like pre-5.0.24, breaks replication from a [5.0.24,5.0.34] master to a fixed (5.0.36) slave! To warn users against this when they upgrade their slave, as agreed with the support team, we add code for a fixed slave to detect that it is connected to a buggy master in a situation (INSERT ON DUP KEY UPDATE into autoinc column) likely to break replication, in which case it cannot replicate so stops and prints a message to the slave's error log and to SHOW SLAVE STATUS. For 5.0.36->[5.0.24,5.0.34] replication we cannot warn as master does not know the slave's version (but we always recommended to users to have slave at least as new as master). As agreed with support, I'll also ask for an alert to be put into the MySQL Network Monitoring and Advisory Service.
2007-02-08 15:53:14 +01:00
return TRUE;
}
}
return FALSE;
}
BUG#33029 5.0 to 5.1 replication fails on dup key when inserting using a trig in SP For all 5.0 and up to 5.1.12 exclusive, when a stored routine or trigger caused an INSERT into an AUTO_INCREMENT column, the generated AUTO_INCREMENT value should not be written into the binary log, which means if a statement does not generate AUTO_INCREMENT value itself, there will be no Intvar event (SET INSERT_ID) associated with it even if one of the stored routine or trigger caused generation of such a value. And meanwhile, when executing a stored routine or trigger, it would ignore the INSERT_ID value even if there is a INSERT_ID value available set by a SET INSERT_ID statement. Starting from MySQL 5.1.12, the generated AUTO_INCREMENT value is written into the binary log, and the value will be used if available when executing the stored routine or trigger. Prior fix of this bug in MySQL 5.0 and prior MySQL 5.1.12 (referenced as the buggy versions in the text below), when a statement that generates AUTO_INCREMENT value by the top statement was executed in the body of a SP, all statements in the SP after this statement would be treated as if they had generated AUTO_INCREMENT by the top statement. When a statement that did not generate AUTO_INCREMENT value by the top statement but by a function/trigger called by it, an erroneous Intvar event would be associated with the statement, this erroneous INSERT_ID value wouldn't cause problem when replicating between masters and slaves of 5.0.x or prior 5.1.12, because the erroneous INSERT_ID value was not used when executing functions/triggers. But when replicating from buggy versions to 5.1.12 or newer, which will use the INSERT_ID value in functions/triggers, the erroneous value will be used, which would cause duplicate entry error and cause the slave to stop. The patch for 5.1 fixed it to ignore the SET INSERT_ID value when executing functions/triggers if it is replicating from a master of buggy versions, another patch for 5.0 fixed it not to generate the erroneous Intvar event.
2008-03-14 04:35:41 +01:00
/**
BUG#33029, For all 5.0 up to 5.0.58 exclusive, and 5.1 up to 5.1.12
exclusive, if one statement in a SP generated AUTO_INCREMENT value
by the top statement, all statements after it would be considered
generated AUTO_INCREMENT value by the top statement, and a
erroneous INSERT_ID value might be associated with these statement,
which could cause duplicate entry error and stop the slave.
Detect buggy master to work around.
*/
bool rpl_master_erroneous_autoinc(THD *thd)
{
if (active_mi && active_mi->rli.sql_thd == thd)
{
Relay_log_info *rli= &active_mi->rli;
Bug#36443 Server crashes when executing insert when insert trigger on table The crash appeared to be a result of allocating an instance of Discrete_interval automatically that that was referred in out-of-declaration scope. Fixed with correcting backing up and restoring scheme of auto_inc_intervals_forced, introduced by bug#33029, by means of shallow copying; added simulation code that forces executing those fixes of the former bug that targeted at master-and-slave having incompatible bug#33029-prone versions.
2008-06-19 20:47:59 +02:00
DBUG_EXECUTE_IF("simulate_bug33029", return TRUE;);
BUG#37426: RBR breaks for CHAR() UTF-8 fields > 85 chars In order to handle CHAR() fields, 8 bits were reserved for the size of the CHAR field. However, instead of denoting the number of characters in the field, field_length was used which denotes the number of bytes in the field. Since UTF-8 fields can have three bytes per character (and has been extended to have four bytes per character in 6.0), an extra two bits have been encoded in the field metadata work for fields of type Field_string (i.e., CHAR fields). Since the metadata word is filled, the extra bits have been encoded in the upper 4 bits of the real type (the most significant byte of the metadata word) by computing the bitwise xor of the extra two bits. Since the upper 4 bits of the real type always is 1111 for Field_string, this means that for fields of length <256, the encoding is identical to the encoding used in pre-5.1.26 servers, but for lengths of 256 or more, an unrecognized type is formed, causing an old slave (that does not handle lengths of 256 or more) to stop.
2008-06-30 22:11:18 +02:00
return rpl_master_has_bug(rli, 33029, FALSE, NULL, NULL);
BUG#33029 5.0 to 5.1 replication fails on dup key when inserting using a trig in SP For all 5.0 and up to 5.1.12 exclusive, when a stored routine or trigger caused an INSERT into an AUTO_INCREMENT column, the generated AUTO_INCREMENT value should not be written into the binary log, which means if a statement does not generate AUTO_INCREMENT value itself, there will be no Intvar event (SET INSERT_ID) associated with it even if one of the stored routine or trigger caused generation of such a value. And meanwhile, when executing a stored routine or trigger, it would ignore the INSERT_ID value even if there is a INSERT_ID value available set by a SET INSERT_ID statement. Starting from MySQL 5.1.12, the generated AUTO_INCREMENT value is written into the binary log, and the value will be used if available when executing the stored routine or trigger. Prior fix of this bug in MySQL 5.0 and prior MySQL 5.1.12 (referenced as the buggy versions in the text below), when a statement that generates AUTO_INCREMENT value by the top statement was executed in the body of a SP, all statements in the SP after this statement would be treated as if they had generated AUTO_INCREMENT by the top statement. When a statement that did not generate AUTO_INCREMENT value by the top statement but by a function/trigger called by it, an erroneous Intvar event would be associated with the statement, this erroneous INSERT_ID value wouldn't cause problem when replicating between masters and slaves of 5.0.x or prior 5.1.12, because the erroneous INSERT_ID value was not used when executing functions/triggers. But when replicating from buggy versions to 5.1.12 or newer, which will use the INSERT_ID value in functions/triggers, the erroneous value will be used, which would cause duplicate entry error and cause the slave to stop. The patch for 5.1 fixed it to ignore the SET INSERT_ID value when executing functions/triggers if it is replicating from a master of buggy versions, another patch for 5.0 fixed it not to generate the erroneous Intvar event.
2008-03-14 04:35:41 +01:00
}
return FALSE;
}
WL#2286 - Compile MySQL w/YASSL support Fix GCC 4.0 link failure. Better CXX_VERSION guessing.
2005-06-22 11:08:28 +02:00
#ifdef HAVE_EXPLICIT_TEMPLATE_INSTANTIATION
Import changeset
2000-07-31 21:29:14 +02:00
template class I_List_iterator<i_string>;
sql/slave.cc buffer size fix + template instatiation
2000-10-21 20:17:08 +02:00
template class I_List_iterator<i_string_pair>;
Import changeset
2000-07-31 21:29:14 +02:00
#endif
Big purge about embedded library (scrum)
2002-12-16 14:33:29 +01:00
BUG#32407: Impossible to do point-in-time recovery from older binlog Problem: it is unsafe to read base64-printed events without first reading the Format_description_log_event (FD). Currently, mysqlbinlog cannot print the FD. As a side effect, another bug has also been fixed: When mysqlbinlog --start-position=X was specified, no ROLLBACK was printed. I changed this, so that ROLLBACK is always printed. This patch does several things: - Format_description_log_event (FD) now print themselves in base64 format. - mysqlbinlog is now able to print FD events. It has three modes: --base64-output=auto Print row events in base64 output, and print FD event. The FD event is printed even if it is outside the range specified with --start-position, because it would not be safe to read row events otherwise. This is the default. --base64-output=always Like --base64-output=auto, but also print base64 output for query events. This is like the old --base64-output flag, which is also a shorthand for --base64-output=always --base64-output=never Never print base64 output, generate error if row events occur in binlog. This is useful to suppress the FD event in binlogs known not to contain row events (e.g., because BINLOG statement is unsafe, requires root privileges, is not SQL, etc) - the BINLOG statement now handles FD events correctly, by setting the thread's rli's relay log's description_event_for_exec to the loaded event. In fact, executing a BINLOG statement is almost the same as reading an event from a relay log. Before my patch, the code for this was separated (exec_relay_log_event in slave.cc executes events from the relay log, mysql_client_binlog_statement in sql_binlog.cc executes BINLOG statements). I needed to augment mysql_client_binlog_statement to do parts of what exec_relay_log_event does. Hence, I did a small refactoring and moved parts of exec_relay_log_event to a new function, which I named apply_event_and_update_pos. apply_event_and_update_pos is called both from exec_relay_log_event and from mysql_client_binlog_statement. - When a non-FD event is executed in a BINLOG statement, without previously executing a FD event in a BINLOG statement, it generates an error, because that's unsafe. I took a new error code for that: ER_NO_FORMAT_DESCRIPTION_EVENT_BEFORE_BINLOG_STATEMENTS. In order to get a decent error message containing the name of the event, I added the class method char* Log_event::get_type_str(Log_event_type type), which returns a string name for the given Log_event_type. This is just like the existing char* Log_event::get_type_str(), except it is a class method that takes the log event type as parameter. I also added PRE_GA_*_ROWS_LOG_EVENT to Log_event::get_type_str(), so that names of old rows event are properly printed. - When reading an event, I added a check that the event type is known by the current Format_description_log_event. Without this, it may crash on bad input (and I was struck by this several times). - I patched the following test cases, which all contain BINLOG statements for row events which must be preceded by BINLOG statements for FD events: - rpl_bug31076 While I was here, I fixed some small things in log_event.cc: - replaced hard-coded 4 by EVENT_TYPE_OFFSET in 3 places - replaced return by DBUG_VOID_RETURN in one place - The name of the logfile can be '-' to indicate stdin. Before my patch, the code just checked if the first character is '-'; now it does a full strcmp(). Probably, all arguments that begin with a - are already handled somewhere else as flags, but I still think it is better that the code reflects what it is supposed to do, with as little dependencies as possible on other parts of the code. If we one day implement that all command line arguments after -- are files (as most unix tools do), then we need this. I also fixed the following in slave.cc: - next_event() was declared twice, and queue_event was not static but should be static (not used outside the file).
2007-12-14 19:02:02 +01:00
/**
@} (end of group Replication)
*/
SCRUM embedded library trimming
2003-01-15 09:11:44 +01:00
#endif /* HAVE_REPLICATION */
Reference in a new issue Copy permalink