Problem:
========
This patch addresses two issues.
First, if a CHANGE MASTER command is issued and an error happens
while locating the replica’s relay logs, the logs can be put into an
invalid state where future updates fail and future CHANGE MASTER
calls crash the server. More specifically, right before a replica
purges the relay logs (part of the `CHANGE MASTER TO` logic), the
relay log is temporarily closed with state LOG_TO_BE_OPENED. If the
server errors in-between the temporary log closure and purge, i.e.
during the function find_log_pos, the log should be closed.
MDEV-25284 reveals the log is not properly closed.
Second, upon issuing a RESET SLAVE ALL command, a slave’s GTID
filters are not cleared (DO_DOMAIN_IDS, IGNORE_DOMIAN_IDS,
IGNORE_SERVER_IDS). MySQL had a similar bug report, Bug #18816897,
which fixed this issue to clear IGNORE_SERVER_IDS after issuing
RESET SLAVE ALL in version 5.7.
Solution:
=========
To fix the first problem, the CHANGE MASTER error handling logic was
extended to transition the relay log state to LOG_CLOSED from
LOG_TO_BE_OPENED.
To fix the second problem, the RESET SLAVE ALL logic is extended to
clear the domain_id filter and ignore_server_ids.
Reviewed By:
============
Andrei Elkin <andrei.elkin@mariadb.com>
In this commit we are adding three more status variable to SHOW SLAVE
STATUS. Slave_DDL_Events and Slave_Non_Transactional_Events.
Slave_DDL_Groups:- This status variable counts the occurrence of DDL
statements
Slave_Non_Transactional_Groups:- This variable count the occurrence
of non-transnational event group.
Slave_Transactional_Groups:- This variable count the occurrence
of transnational event group.
Patch Credit:- Kristian Nielsen
commit 3dc3ab1a30 introduced
Rows_event_tracker, using a mismatch of size_t (the native
register width) and my_off_t (the file offset width, usually
64 bits). Use my_off_t both in member fields and member functions.
Problems --------
The slave io thread did not conduct integrity check
for a group of row-based events. Specifically it tolerates missed
terminal block event that must be flagged with STMT_END. Failure to
react on its loss can confuse the applier thread in various ways.
Another potential issue was that there were no check of impossible
second in row Gtid-log-event while the slave io thread is receiving
to be skipped events after reconnect.
Fixes
-----
The slave io thread is made by this patch to track the rows event
STMT_END status.
Whenever at next event reading the IO thread finds out that a preceding
Rows event did not actually had the flag, an
explicit error is issued.
Replication can be resumed after the source of failure is eliminated,
see a provided test.
Note that currently the row-based group integrity check excludes
the compressed version 2 Rows events (which are not generated by MariaDB
master).
Its uncompressed counterpart is manually tested.
The 2nd issue is covered to produce an error in case the io thread
receives a successive Gtid_log_event while it is post-reconnect
skipping.
and specifically the ack receiving functionality.
Semisync is turned to be static instead of plugin so its functions
are invoked at the same points as RUN_HOOKS.
The RUN_HOOKS and the observer interface remain to be removed by later
patch.
Todo:
React on killed status by repl_semisync_master.wait_after_sync(). Currently
Repl_semi_sync_master::commit_trx does not check the killed status.
There were few bugfixes found that are present in mysql and its unclear
whether/how they are covered. Those include:
Bug#15985893: GTID SKIPPED EVENTS ON MASTER CAUSE SEMI SYNC TIME-OUTS
Bug#17932935 CALLING IS_SEMI_SYNC_SLAVE() IN EACH FUNCTION CALL
HAS BAD PERFORMANCE
Bug#20574628: SEMI-SYNC REPLICATION PERFORMANCE DEGRADES WITH A HIGH NUMBER OF THREADS
- Added sql/mariadb.h file that should be included first by files in sql
directory, if sql_plugin.h is not used (sql_plugin.h adds SHOW variables
that must be done before my_global.h is included)
- Removed a lot of include my_global.h from include files
- Removed include's of some files that my_global.h automatically includes
- Removed duplicated include's of my_sys.h
- Replaced include my_config.h with my_global.h
Problem:- In the case of multisource replication/named slave
when we run "FLUSH LOGS" , it does not flush logs.
Solution:- A new function Master_info_index->flush_all_relay_logs()
is created which will rotate relay logs for all named slave.
This will be called in reload_acl_and_cache function when
connection_name.length == 0
Benefits of this patch:
- Removed a lot of calls to strlen(), especially for field_string
- Strings generated by parser are now const strings, less chance of
accidently changing a string
- Removed a lot of calls with LEX_STRING as parameter (changed to pointer)
- More uniform code
- Item::name_length was not kept up to date. Now fixed
- Several bugs found and fixed (Access to null pointers,
access of freed memory, wrong arguments to printf like functions)
- Removed a lot of casts from (const char*) to (char*)
Changes:
- This caused some ABI changes
- lex_string_set now uses LEX_CSTRING
- Some fucntions are now taking const char* instead of char*
- Create_field::change and after changed to LEX_CSTRING
- handler::connect_string, comment and engine_name() changed to LEX_CSTRING
- Checked printf() related calls to find bugs. Found and fixed several
errors in old code.
- A lot of changes from LEX_STRING to LEX_CSTRING, especially related to
parsing and events.
- Some changes from LEX_STRING and LEX_STRING & to LEX_CSTRING*
- Some changes for char* to const char*
- Added printf argument checking for my_snprintf()
- Introduced null_clex_str, star_clex_string, temp_lex_str to simplify
code
- Added item_empty_name and item_used_name to be able to distingush between
items that was given an empty name and items that was not given a name
This is used in sql_yacc.yy to know when to give an item a name.
- select table_name."*' is not anymore same as table_name.*
- removed not used function Item::rename()
- Added comparision of item->name_length before some calls to
my_strcasecmp() to speed up comparison
- Moved Item_sp_variable::make_field() from item.h to item.cc
- Some minimal code changes to avoid copying to const char *
- Fixed wrong error message in wsrep_mysql_parse()
- Fixed wrong code in find_field_in_natural_join() where real_item() was
set when it shouldn't
- ER_ERROR_ON_RENAME was used with extra arguments.
- Removed some (wrong) ER_OUTOFMEMORY, as alloc_root will already
give the error.
TODO:
- Check possible unsafe casts in plugin/auth_examples/qa_auth_interface.c
- Change code to not modify LEX_CSTRING for database name
(as part of lower_case_table_names)
Intermediate commit.
Move the discovery of mysql.gtid_slave_pos* tables into the SQL thread.
This avoids doing things like opening tables and scanning the mysql
schema for tables inside of the START SLAVE statement, which might
interact badly with existing transaction or table locks.
(Even though START SLAVE is documented to implicitly commit any active
transactions, this appears not to be the case in current code).
Table discovery fits naturally in the SQL thread init code, next to
the loading of mysql.gtid_slave_pos state.
The reason for this is that stop slave takes LOCK_active_mi over the
whole operation while some slave operations will also need LOCK_active_mi
which causes deadlocks.
Fixed by introducing object counting for Master_info and not taking
LOCK_active_mi over stop slave or even stop_all_slaves()
Another benefit of this approach is that it allows:
- Multiple threads can run SHOW SLAVE STATUS at the same time
- START/STOP/RESET/SLAVE STATUS on a slave will not block other slaves
- Simpler interface for handling get_master_info()
- Added some missing unlock of 'log_lock' in error condtions
- Moved rpl_parallel_inactivate_pool(&global_rpl_thread_pool) to end
of stop_slave() to not have to use LOCK_active_mi inside
terminate_slave_threads()
- Changed argument for remove_master_info() to Master_info, as we always
have this available
- Fixed core dump when doing FLUSH TABLES WITH READ LOCK and parallel
replication. Problem was that waiting for pause_for_ftwrl was not done
when deleting rpt->current_owner after a force_abort.
This has no functional changes, but it helps avoid merge problems from 10.0
to 10.1. In 10.0, code that checks for parallel replication uses
opt_slave_parallel_threads > 0, but this check needs to be
mi->using_parallel() in 10.1. By using the same check in 10.0 (with
unchanged semantics), merge problems to 10.1 are avoided.
- Added testing if connection is killed to shortcut reading of connection data
This will allow us later in 10.2 to do a cleaner shutdown of slaves (less errors in the log)
- Add new status variables: Slaves_connected, Slaves_running and Slave_connections.
- Use MYSQL_SLAVE_NOT_RUN instead of 0 with slave_running.
- Don't print obvious extra warnings to the error log when slave is shut down normally.
Delay spawning parallel replication worker threads until a slave SQL
thread is running, and de-spawn them when the last SQL thread stops.
This is especially useful to avoid needless threads on a master in a
setup where same my.cnf is used on masters and slaves.
Adjust the configuration options, as discussed on the
maria-developers@ mailing list.
The option to hint a transaction to not be replicated in parallel is
now called @@skip_parallel_replication, consistent with
@@skip_replication.
And the --slave-parallel-mode is now simplified to have just one of
the following values:
none
minimal
conservative
optimistic
aggressive
This reflects successively harder efforts to find opportunities to run
things in parallel on the slave. It allows to extend the server with
more automatic heuristics in the future without having to introduce a
new configuration option for each and every one.
Implement a new mode for parallel replication. In this mode, all transactions
are optimistically attempted applied in parallel. In case of conflicts, the
offending transaction is rolled back and retried later non-parallel.
This is an early-release patch to facilitate testing, more changes to user
interface / options will be expected. The new mode is not enabled by default.
If the slave gets a reconnect in the middle of a GTID event group, normally
it will re-fetch that event group, skipping the first part that was already
queued for the SQL thread.
However, if the master crashed while writing the event group, the group is
incomplete. This patch detects this case and makes sure that the
transaction is rolled back and nothing is skipped from any following
event groups.
Similarly, a network proxy might cause the reconnect to end up on a
different master server. Detect this by noticing a different server_id,
and similarly in this case roll back the partially received group.
The previous patch for this bug was unfortunately completely wrong.
The purpose of cached_charset is to remember which character set we
have installed currently in the THD, so that in the common case where
charset does not change between queries, we do not need to update it
in the THD. Thus, it is important that the cached_charset field is
tightly coupled to the THD for which it handles caching.
Thus the right place to put cached_charset seems to be in the THD.
This patch introduces a field THD:system_thread_info where such info
in general can be placed without further inflating the THD with unused
data for other threads (THD is already far too big as it is). It then
moves the cached_charset into this slot for the SQL driver thread and
for the parallel replication worker threads.
The THD::rpl_filter field is also moved inside system_thread_info, to
keep the size of THD unchanged. Moving further fields in to reduce the
size of THD is a separate task, filed as MDEV-6164.
Make IO thread check for end of event group, so that upon
disconnect at the end of an event group it can report the last
read GTID as expected. Also inject a fake Rotate event at
reconnect when skipping part of an initial event group, to
give SQL thread the correct Read_Master_Log_Pos.
Reported by Pavel Ivanov.
includes:
* remove some remnants of "Bug#14521864: MYSQL 5.1 TO 5.5 BUGS PARTITIONING"
* introduce LOCK_share, now LOCK_ha_data is strictly for engines
* rea_create_table() always creates .par file (even in "frm-only" mode)
* fix a 5.6 bug, temp file leak on dummy ALTER TABLE
Fix problems related to reconnect. When we need to reconnect (ie. explict
stop/start of just the IO thread by user, or automatic reconnect due to
loosing network connection with the master), it is a bit complex to correctly
resume at the right point without causing duplicate or missing events in the
relay log. The previous code had multiple problems in this regard.
With this patch, the problem is solved as follows. The IO thread keeps track
(in memory) of which GTID was last queued to the relay log. If it needs to
reconnect, it resumes at that GTID position. It also counts number of events
received within the last, possibly partial, event group, and skips the same
number of events after a reconnect, so that events already enqueued before the
reconnect are not duplicated.
(There is no need to keep any persistent state; whenever we restart slave
threads after both of them being stopped (such as after server restart), we
erase the relay logs and start over from the last GTID applied by SQL thread.
But while the SQL thread is running, this patch is needed to get correct relay
log).
Change of user interface to be more logical and more in line with expectations
to work similar to old-style replication.
User can now explicitly choose in CHANGE MASTER whether binlog position is
taken into account (master_gtid_pos=current_pos) or not (master_gtid_pos=
slave_pos) when slave connects to master.
@@gtid_pos is replaced by three separate variables @@gtid_slave_pos (can
be set by user, replicated GTIDs only), @@gtid_binlog_pos (read only), and
@@gtid_current_pos (a combination of the two, most recent GTID within each
domain). mysql.rpl_slave_state is renamed to mysql.gtid_slave_pos to match.
This fixes MDEV-4474.
Users can set different repplication filter rules for each replication connection, in my.cnf or command line.
But the rules set online will not record in master.info, it means if users restart MySQL, these rules will lose.
So if users wantn't their replication filter rules lose, they should write the rules in my.cnf.
Users can set rules by 2 ways:
1. Online SET command, "SET connection_name.replication_filter_settings = rules;".
2. In my.cnf, "connection_name.replication_filter_settings = rules".
If no connection_name in my.cnf, this rule will apply for ALL replication connection.
If no connetion_name in SET statement, this rull will apply for default_connection_name.
Fix MDEV-4275 - I/O thread restart duplicates events in the relay log.
The first time we connect to master after CHANGE MASTER or restart, we connect
from the GTID position. But then subsequent reconnects or IO thread restarts
reconnect with the old-style file/offset binlog pos from where it left off at
last disconnect. This is necessary to avoid duplicate events in the relay
logs, as there is nothing that synchronises the SQL thread update of GTID
state (multiple threads in case of multi-source) with IO thread reconnects.
Test cases.
Some small cleanups and fixes.
