mirror/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2025-01-18 13:02:28 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions
mariadb/sql/log_event_old.h

572 lines
20 KiB
C
Raw Normal View History

WL#2735: Refactor replication Factoring out pack_row() and unpack_row() into new files rpl_record.{cc,h}.
2007-04-13 14:55:28 +02:00
/* Copyright 2007 MySQL AB. All rights reserved.
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
the Free Software Foundation; version 2 of the License.
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.
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#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
#ifndef LOG_EVENT_OLD_H
#define LOG_EVENT_OLD_H
/*
Need to include this file at the proper position of log_event.h
*/
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
/**
@file
@brief This file contains classes handling old formats of row-based
binlog events.
*/
/*
Around 2007-10-31, I made these classes completely separated from
the new classes (before, there was a complex class hierarchy
involving multiple inheritance; see BUG#31581), by simply copying
and pasting the entire contents of Rows_log_event into
Old_rows_log_event and the entire contents of
{Write|Update|Delete}_rows_log_event into
{Write|Update|Delete}_rows_log_event_old. For clarity, I will keep
the comments marking which code was cut-and-pasted for some time.
With the classes collapsed into one, there is probably some
redundancy (maybe some methods can be simplified and/or removed),
but we keep them this way for now. /Sven
*/
/**
@class Old_rows_log_event
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
Base class for the three types of row-based events
{Write|Update|Delete}_row_log_event_old, with event type codes
PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. These events are never
created any more, except when reading a relay log created by an old
server.
*/
class Old_rows_log_event : public Log_event
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
{
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
/********** BEGIN CUT & PASTE FROM Rows_log_event **********/
public:
/**
Enumeration of the errors that can be returned.
*/
enum enum_error
{
ERR_OPEN_FAILURE = -1, /**< Failure to open table */
ERR_OK = 0, /**< No error */
ERR_TABLE_LIMIT_EXCEEDED = 1, /**< No more room for tables */
ERR_OUT_OF_MEM = 2, /**< Out of memory */
ERR_BAD_TABLE_DEF = 3, /**< Table definition does not match */
ERR_RBR_TO_SBR = 4 /**< daisy-chanining RBR to SBR not allowed */
};
/*
These definitions allow you to combine the flags into an
appropriate flag set using the normal bitwise operators. The
implicit conversion from an enum-constant to an integer is
accepted by the compiler, which is then used to set the real set
of flags.
*/
enum enum_flag
{
/* Last event of a statement */
STMT_END_F = (1U << 0),
/* Value of the OPTION_NO_FOREIGN_KEY_CHECKS flag in thd->options */
NO_FOREIGN_KEY_CHECKS_F = (1U << 1),
/* Value of the OPTION_RELAXED_UNIQUE_CHECKS flag in thd->options */
RELAXED_UNIQUE_CHECKS_F = (1U << 2),
/**
Indicates that rows in this event are complete, that is contain
values for all columns of the table.
*/
COMPLETE_ROWS_F = (1U << 3)
};
typedef uint16 flag_set;
/* Special constants representing sets of flags */
enum
{
RLE_NO_FLAGS = 0U
};
virtual ~Old_rows_log_event();
void set_flags(flag_set flags_arg) { m_flags |= flags_arg; }
void clear_flags(flag_set flags_arg) { m_flags &= ~flags_arg; }
flag_set get_flags(flag_set flags_arg) const { return m_flags & flags_arg; }
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
virtual void pack_info(Protocol *protocol);
#endif
#ifdef MYSQL_CLIENT
/* not for direct call, each derived has its own ::print() */
virtual void print(FILE *file, PRINT_EVENT_INFO *print_event_info)= 0;
#endif
#ifndef MYSQL_CLIENT
int add_row_data(uchar *data, size_t length)
{
return do_add_row_data(data,length);
}
#endif
/* Member functions to implement superclass interface */
virtual int get_data_size();
MY_BITMAP const *get_cols() const { return &m_cols; }
size_t get_width() const { return m_width; }
ulong get_table_id() const { return m_table_id; }
#ifndef MYSQL_CLIENT
virtual bool write_data_header(IO_CACHE *file);
virtual bool write_data_body(IO_CACHE *file);
virtual const char *get_db() { return m_table->s->db.str; }
#endif
/*
Check that malloc() succeeded in allocating memory for the rows
buffer and the COLS vector. Checking that an Update_rows_log_event_old
is valid is done in the Update_rows_log_event_old::is_valid()
function.
*/
virtual bool is_valid() const
{
return m_rows_buf && m_cols.bitmap;
}
uint m_row_count; /* The number of rows added to the event */
protected:
/*
The constructors are protected since you're supposed to inherit
this class, not create instances of this class.
*/
#ifndef MYSQL_CLIENT
Old_rows_log_event(THD*, TABLE*, ulong table_id,
MY_BITMAP const *cols, bool is_transactional);
#endif
Old_rows_log_event(const char *row_data, uint event_len,
Log_event_type event_type,
const Format_description_log_event *description_event);
#ifdef MYSQL_CLIENT
void print_helper(FILE *, PRINT_EVENT_INFO *, char const *const name);
#endif
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
#ifndef MYSQL_CLIENT
virtual int do_add_row_data(uchar *data, size_t length);
#endif
#ifndef MYSQL_CLIENT
TABLE *m_table; /* The table the rows belong to */
#endif
ulong m_table_id; /* Table ID */
MY_BITMAP m_cols; /* Bitmap denoting columns available */
ulong m_width; /* The width of the columns bitmap */
ulong m_master_reclength; /* Length of record on master side */
/* Bit buffers in the same memory as the class */
uint32 m_bitbuf[128/(sizeof(uint32)*8)];
uint32 m_bitbuf_ai[128/(sizeof(uint32)*8)];
uchar *m_rows_buf; /* The rows in packed format */
uchar *m_rows_cur; /* One-after the end of the data */
uchar *m_rows_end; /* One-after the end of the allocated space */
flag_set m_flags; /* Flags for row-level events */
/* helper functions */
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
const uchar *m_curr_row; /* Start of the row being processed */
const uchar *m_curr_row_end; /* One-after the end of the current row */
uchar *m_key; /* Buffer to keep key value during searches */
int find_row(const Relay_log_info *const);
int write_row(const Relay_log_info *const, const bool);
// Unpack the current row into m_table->record[0]
int unpack_current_row(const Relay_log_info *const rli)
{
DBUG_ASSERT(m_table);
ASSERT_OR_RETURN_ERROR(m_curr_row < m_rows_end, HA_ERR_CORRUPT_EVENT);
int const result= ::unpack_row(rli, m_table, m_width, m_curr_row, &m_cols,
&m_curr_row_end, &m_master_reclength);
ASSERT_OR_RETURN_ERROR(m_curr_row_end <= m_rows_end, HA_ERR_CORRUPT_EVENT);
return result;
}
#endif
private:
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
virtual int do_apply_event(Relay_log_info const *rli);
virtual int do_update_pos(Relay_log_info *rli);
virtual enum_skip_reason do_shall_skip(Relay_log_info *rli);
/*
Primitive to prepare for a sequence of row executions.
DESCRIPTION
Before doing a sequence of do_prepare_row() and do_exec_row()
calls, this member function should be called to prepare for the
entire sequence. Typically, this member function will allocate
space for any buffers that are needed for the two member
functions mentioned above.
RETURN VALUE
The member function will return 0 if all went OK, or a non-zero
error code otherwise.
*/
virtual
int do_before_row_operations(const Slave_reporting_capability *const log) = 0;
/*
Primitive to clean up after a sequence of row executions.
DESCRIPTION
After doing a sequence of do_prepare_row() and do_exec_row(),
this member function should be called to clean up and release
any allocated buffers.
The error argument, if non-zero, indicates an error which happened during
row processing before this function was called. In this case, even if
function is successful, it should return the error code given in the argument.
*/
virtual
int do_after_row_operations(const Slave_reporting_capability *const log,
int error) = 0;
/*
Primitive to do the actual execution necessary for a row.
DESCRIPTION
The member function will do the actual execution needed to handle a row.
The row is located at m_curr_row. When the function returns,
m_curr_row_end should point at the next row (one byte after the end
of the current row).
RETURN VALUE
0 if execution succeeded, 1 if execution failed.
*/
virtual int do_exec_row(const Relay_log_info *const rli) = 0;
#endif /* !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION) */
/********** END OF CUT & PASTE FROM Rows_log_event **********/
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
protected:
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
int do_apply_event(Old_rows_log_event*,const Relay_log_info*);
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
/*
Primitive to prepare for a sequence of row executions.
DESCRIPTION
Before doing a sequence of do_prepare_row() and do_exec_row()
calls, this member function should be called to prepare for the
entire sequence. Typically, this member function will allocate
space for any buffers that are needed for the two member
functions mentioned above.
RETURN VALUE
The member function will return 0 if all went OK, or a non-zero
error code otherwise.
*/
virtual int do_before_row_operations(TABLE *table) = 0;
/*
Primitive to clean up after a sequence of row executions.
DESCRIPTION
After doing a sequence of do_prepare_row() and do_exec_row(),
this member function should be called to clean up and release
any allocated buffers.
*/
virtual int do_after_row_operations(TABLE *table, int error) = 0;
/*
Primitive to prepare for handling one row in a row-level event.
DESCRIPTION
The member function prepares for execution of operations needed for one
row in a row-level event by reading up data from the buffer containing
the row. No specific interpretation of the data is normally done here,
since SQL thread specific data is not available: that data is made
available for the do_exec function.
A pointer to the start of the next row, or NULL if the preparation
failed. Currently, preparation cannot fail, but don't rely on this
behavior.
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
RETURN VALUE
Error code, if something went wrong, 0 otherwise.
*/
Merge sita.local:/Users/tsmith/m/bk/maint/51-target22 into sita.local:/Users/tsmith/m/bk/maint/51
2007-08-29 23:28:38 +02:00
virtual int do_prepare_row(THD*, Relay_log_info const*, TABLE*,
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
uchar const *row_start,
uchar const **row_end) = 0;
/*
Primitive to do the actual execution necessary for a row.
DESCRIPTION
The member function will do the actual execution needed to handle a row.
RETURN VALUE
0 if execution succeeded, 1 if execution failed.
*/
virtual int do_exec_row(TABLE *table) = 0;
#endif /* !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION) */
};
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
/**
@class Write_rows_log_event_old
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
Old class for binlog events that write new rows to a table (event
type code PRE_GA_WRITE_ROWS_EVENT). Such events are never produced
by this version of the server, but they may be read from a relay log
created by an old server. New servers create events of class
Write_rows_log_event (event type code WRITE_ROWS_EVENT) instead.
*/
class Write_rows_log_event_old : public Old_rows_log_event
{
/********** BEGIN CUT & PASTE FROM Write_rows_log_event **********/
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
public:
Fixing compile problem on some platforms resulting from building a configuration with HAVE_REPLICATION undefined, which causes the *_rows_log_event_old classes to only have a default constructor. This generates an error on some platforms, since the base classes does not have a default constructor defined. Updating signatures of do_prepare_row() for classes above, since signatures in base classes changed.
2007-04-13 11:19:31 +02:00
#if !defined(MYSQL_CLIENT)
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
Write_rows_log_event_old(THD*, TABLE*, ulong table_id,
MY_BITMAP const *cols, bool is_transactional);
Fixing compile problem on some platforms resulting from building a configuration with HAVE_REPLICATION undefined, which causes the *_rows_log_event_old classes to only have a default constructor. This generates an error on some platforms, since the base classes does not have a default constructor defined. Updating signatures of do_prepare_row() for classes above, since signatures in base classes changed.
2007-04-13 11:19:31 +02:00
#endif
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
#ifdef HAVE_REPLICATION
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
Write_rows_log_event_old(const char *buf, uint event_len,
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
const Format_description_log_event *description_event);
#endif
#if !defined(MYSQL_CLIENT)
static bool binlog_row_logging_function(THD *thd, TABLE *table,
bool is_transactional,
MY_BITMAP *cols,
uint fields,
const uchar *before_record
__attribute__((unused)),
const uchar *after_record)
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
{
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
return thd->binlog_write_row(table, is_transactional,
cols, fields, after_record);
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
}
#endif
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
private:
#ifdef MYSQL_CLIENT
void print(FILE *file, PRINT_EVENT_INFO *print_event_info);
#endif
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
virtual int do_before_row_operations(const Slave_reporting_capability *const);
virtual int do_after_row_operations(const Slave_reporting_capability *const,int);
virtual int do_exec_row(const Relay_log_info *const);
#endif
/********** END OF CUT & PASTE FROM Write_rows_log_event **********/
public:
enum
{
/* Support interface to THD::binlog_prepare_pending_rows_event */
TYPE_CODE = PRE_GA_WRITE_ROWS_EVENT
};
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
private:
virtual Log_event_type get_type_code() { return (Log_event_type)TYPE_CODE; }
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
// use old definition of do_apply_event()
Merge sita.local:/Users/tsmith/m/bk/maint/51-target22 into sita.local:/Users/tsmith/m/bk/maint/51
2007-08-29 23:28:38 +02:00
virtual int do_apply_event(const Relay_log_info *rli)
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
{ return Old_rows_log_event::do_apply_event(this,rli); }
// primitives for old version of do_apply_event()
virtual int do_before_row_operations(TABLE *table);
virtual int do_after_row_operations(TABLE *table, int error);
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
virtual int do_prepare_row(THD*, Relay_log_info const*, TABLE*,
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 const *row_start, uchar const **row_end);
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
virtual int do_exec_row(TABLE *table);
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
#endif
};
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
/**
@class Update_rows_log_event_old
Old class for binlog events that modify existing rows to a table
(event type code PRE_GA_UPDATE_ROWS_EVENT). Such events are never
produced by this version of the server, but they may be read from a
relay log created by an old server. New servers create events of
class Update_rows_log_event (event type code UPDATE_ROWS_EVENT)
instead.
*/
class Update_rows_log_event_old : public Old_rows_log_event
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
{
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
/********** BEGIN CUT & PASTE FROM Update_rows_log_event **********/
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
public:
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
#ifndef MYSQL_CLIENT
Update_rows_log_event_old(THD*, TABLE*, ulong table_id,
MY_BITMAP const *cols,
bool is_transactional);
#endif
#ifdef HAVE_REPLICATION
Update_rows_log_event_old(const char *buf, uint event_len,
const Format_description_log_event *description_event);
#endif
#if !defined(MYSQL_CLIENT)
static bool binlog_row_logging_function(THD *thd, TABLE *table,
bool is_transactional,
MY_BITMAP *cols,
uint fields,
const uchar *before_record,
const uchar *after_record)
Fixing compile problem on some platforms resulting from building a configuration with HAVE_REPLICATION undefined, which causes the *_rows_log_event_old classes to only have a default constructor. This generates an error on some platforms, since the base classes does not have a default constructor defined. Updating signatures of do_prepare_row() for classes above, since signatures in base classes changed.
2007-04-13 11:19:31 +02:00
{
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
return thd->binlog_update_row(table, is_transactional,
cols, fields, before_record, after_record);
Fixing compile problem on some platforms resulting from building a configuration with HAVE_REPLICATION undefined, which causes the *_rows_log_event_old classes to only have a default constructor. This generates an error on some platforms, since the base classes does not have a default constructor defined. Updating signatures of do_prepare_row() for classes above, since signatures in base classes changed.
2007-04-13 11:19:31 +02:00
}
#endif
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
protected:
#ifdef MYSQL_CLIENT
void print(FILE *file, PRINT_EVENT_INFO *print_event_info);
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
#endif
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
virtual int do_before_row_operations(const Slave_reporting_capability *const);
virtual int do_after_row_operations(const Slave_reporting_capability *const,int);
virtual int do_exec_row(const Relay_log_info *const);
#endif /* !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION) */
/********** END OF CUT & PASTE FROM Update_rows_log_event **********/
uchar *m_after_image, *m_memory;
public:
enum
{
/* Support interface to THD::binlog_prepare_pending_rows_event */
TYPE_CODE = PRE_GA_UPDATE_ROWS_EVENT
};
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
private:
virtual Log_event_type get_type_code() { return (Log_event_type)TYPE_CODE; }
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
// use old definition of do_apply_event()
Merge sita.local:/Users/tsmith/m/bk/maint/51-target22 into sita.local:/Users/tsmith/m/bk/maint/51
2007-08-29 23:28:38 +02:00
virtual int do_apply_event(const Relay_log_info *rli)
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
{ return Old_rows_log_event::do_apply_event(this,rli); }
// primitives for old version of do_apply_event()
virtual int do_before_row_operations(TABLE *table);
virtual int do_after_row_operations(TABLE *table, int error);
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
virtual int do_prepare_row(THD*, Relay_log_info const*, TABLE*,
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 const *row_start, uchar const **row_end);
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
virtual int do_exec_row(TABLE *table);
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
#endif /* !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION) */
};
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
/**
@class Delete_rows_log_event_old
Old class for binlog events that delete existing rows from a table
(event type code PRE_GA_DELETE_ROWS_EVENT). Such events are never
produced by this version of the server, but they may be read from a
relay log created by an old server. New servers create events of
class Delete_rows_log_event (event type code DELETE_ROWS_EVENT)
instead.
*/
class Delete_rows_log_event_old : public Old_rows_log_event
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
{
BUG#31581: 5.1-telco-6.1 -> 5.1.22. Slave crashes during starting This patch has two purposes: (1) To refactor the code so that {Write|Update|Delete}_rows_log_event_old does not use code from {Write|Update|Delete}_rows_log_event. Before refactoring there was the following problem: whenever we modifed the code for new events, it affected the old events. This is bad, as it makes maintainance difficult. After refactoring, we can safely edit the new code without affecting old events. So, if we for instance modify the binary format of new events, we no longer need to worry about how the new code reads old events. (2) To fix BUG#31581. These two objectives are reached by the following changes: - Merged Rows_log_event into Old_rows_log_event and {Write|Update|Delete}_rows_log_event into {Write|Update|Delete}_rows_log_event_old. - Fixed the bug by replacing {WRITE|UPDATE|DELETE}_ROWS_EVENT by PRE_GA_{WRITE|UPDATE|DELETE}_ROWS_EVENT. - Added comments to log_event_old.h (This patch is identical to the previously committed patch which was a collapse of three changesets, except that it adds assert(0) to constructors for old types of row log events that should never be called.)
2007-11-20 19:49:35 +01:00
/********** BEGIN CUT & PASTE FROM Update_rows_log_event **********/
public:
#ifndef MYSQL_CLIENT
Delete_rows_log_event_old(THD*, TABLE*, ulong,
MY_BITMAP const *cols, bool is_transactional);
#endif
#ifdef HAVE_REPLICATION
Delete_rows_log_event_old(const char *buf, uint event_len,
const Format_description_log_event *description_event);
#endif
#if !defined(MYSQL_CLIENT)
static bool binlog_row_logging_function(THD *thd, TABLE *table,
bool is_transactional,
MY_BITMAP *cols,
uint fields,
const uchar *before_record,
const uchar *after_record
__attribute__((unused)))
{
return thd->binlog_delete_row(table, is_transactional,
cols, fields, before_record);
}
#endif
protected:
#ifdef MYSQL_CLIENT
void print(FILE *file, PRINT_EVENT_INFO *print_event_info);
#endif
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
virtual int do_before_row_operations(const Slave_reporting_capability *const);
virtual int do_after_row_operations(const Slave_reporting_capability *const,int);
virtual int do_exec_row(const Relay_log_info *const);
#endif
/********** END CUT & PASTE FROM Delete_rows_log_event **********/
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
uchar *m_after_image, *m_memory;
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
public:
enum
{
/* Support interface to THD::binlog_prepare_pending_rows_event */
TYPE_CODE = PRE_GA_DELETE_ROWS_EVENT
};
private:
virtual Log_event_type get_type_code() { return (Log_event_type)TYPE_CODE; }
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
// use old definition of do_apply_event()
Merge sita.local:/Users/tsmith/m/bk/maint/51-target22 into sita.local:/Users/tsmith/m/bk/maint/51
2007-08-29 23:28:38 +02:00
virtual int do_apply_event(const Relay_log_info *rli)
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
{ return Old_rows_log_event::do_apply_event(this,rli); }
// primitives for old version of do_apply_event()
virtual int do_before_row_operations(TABLE *table);
virtual int do_after_row_operations(TABLE *table, int error);
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
virtual int do_prepare_row(THD*, Relay_log_info const*, TABLE*,
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 const *row_start, uchar const **row_end);
BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 using TPC-B): Problem: A RBR event can contain incomplete row data (only key value and fields which have been changed). In that case, when the row is unpacked into record and written to a table, the missing fields get incorrect NULL values leading to master-slave inconsistency. Solution: Use values found in slave's table for columns which are not given in the rows event. The code for writing a single row uses the following algorithm: 1. unpack row_data into table->record[0], 2. try to insert record, 3. if duplicate record found, fetch it into table->record[0], 4. unpack row_data into table->record[0], 5. write table->record[0] into the table. Where row_data is the row as stored in the data area of a rows event. Thus: a) unpacking of row_data happens at the time when row is written into a table, b) when unpacking (in step 4), only columns present in row_data are overwritten - all other columns remain as they were found in the table. Since all data needed for the above algorithm is stored inside Rows_log_event class, functions which locate and write rows are turned into methods of that class. replace_record() -> Rows_log_event::write_row() find_and_fetch_row() -> Rows_log_event::find_row() Both methods take row data from event's data buffer - the row being processed is pointed by m_curr_row. They unpack the data as needed into table's record buffers record[0] or record[1]. When row is unpacked, m_curr_row_end is set to point at next row in the data buffer. Other changes introduced in this changeset: - Change signature of unpack_row(): don't report errors and don't setup table's rw_set here. Errors can happen only when setting default values in prepare_record() function and are detected there. - In Rows_log_event and derived classes, don't pass arguments to the execution primitives (do_...() member functions) but use class members instead. - Move old row handling code into log_event_old.cc to be used by *_rows_log_event_old classes. Also, a new test rpl_ndb_2other is added which tests basic replication from master using ndb tables to slave storing the same tables using (possibly) different engine (myisam,innodb). Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. However, these tests doesn't work for various reasons and currently are disabled (see BUG#19227). The new test differs from the ones it is based on as follows: 1. Single test tests replication with different storage engines on slave (myisam, innodb, ndb). 2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test which doesn't contain tests using partitioned tables as these don't work currently. Instead, it tests replication to a slave which has more or less columns than master. 3. Include file include/rpl_multi_engine3.inc is replaced with include/rpl_multi_engine2.inc. The later differs by performing slightly different operations (updating more than one row in the table) and clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" as replication of "DELETE" doesn't work well in this setting. 4. Slave must use option --log-slave-updates=0 as otherwise execution of replication events generated by ndb fails if table uses a different storage engine on slave (see BUG#29569).
2007-08-26 14:31:10 +02:00
virtual int do_exec_row(TABLE *table);
BUG#27779 (Slave cannot read old rows log events): Taking code from before BUG#22583 and incorporating as events to be able to read old events. Also incorporating old pack and unpack functions into patch.
2007-04-12 15:50:54 +02:00
#endif
};
#endif
Reference in a new issue Copy permalink