This is to get same test results on Linux and Windows
modified:
storage/connect/mysql-test/connect/r/part_file.result
storage/connect/mysql-test/connect/r/part_table.result
storage/connect/mysql-test/connect/t/part_file.test
storage/connect/mysql-test/connect/t/part_table.test
storage/connect/tabdos.cpp
If mysql.innodb_table_stats or mysql.innodb_index_stats is not found or has
unexpected structure output that error only once and no other error for
every table trying to use them. If they do exists, then print fetch or
recalculation errors only once / table or index.
modified:
storage/connect/ha_connect.cc
storage/connect/tabdos.cpp
storage/connect/tabfmt.cpp
storage/connect/tabmysql.cpp
- add AVG_REC_LENGTH option to avoid result mismatch between
Windows and Linux
modified:
storage/connect/mysql-test/connect/r/part_file.result
storage/connect/mysql-test/connect/r/part_table.result
storage/connect/mysql-test/connect/t/part_file.test
line 8473
In case InnoDB index is not found, print the MySQL and InnoDB index
name we were trying to find and all MySQL and InnoDB index names there
is for this table.
ha_innodb.cc line 8473
If index is not found from InnoDB make sure we print what we
were trying to find and all mysql and InnoDB index names there
is for this table.
modified:
storage/connect/array.cpp
storage/connect/array.h
storage/connect/blkfil.cpp
storage/connect/blkfil.h
storage/connect/filter.cpp
storage/connect/filter.h
storage/connect/ha_connect.cc
storage/connect/tabdos.cpp
- FIX problems of DIR table on LINUX:
The order of the result is not the same than the one on Windows
An error occurs when no file match the pattern
Also the row numbers may also be different (not fixed yet)
modified:
storage/connect/mysql-test/connect/r/part_file.result
storage/connect/mysql-test/connect/r/part_table.result
storage/connect/mysql-test/connect/t/part_file.test
a sub-branch of this one and merged by pushing all the changes from it.
This version adds the following to CONNECT:
- MRR support (similar to the MyISAM one)
- Block, Remote and dynamic indexing
- Partitioning support (using the PARTITION engine)
Here is a list of the commited changes made in the sub-branch:
========================================================================
------------------------------------------------------------
revno: 4009
committer: Olivier Bertrand <bertrandop@gmail.com>
branch nick: 10.0-connect
timestamp: Thu 2014-07-17 18:13:51 +0200
message:
This commit brings many changes, in particular two important ones:
1) Support of partitioning by connect. A table can be partitioned
by files, this is an enhanced MULTIPLE table. It can be also
partitioned by sub-tables like TBL and this enables table sharding.
2) Handling a CONNECT bug that causes in some cases extraneous rows
to remain in the table after an UPDATE or DELETE when the command
uses indexing (for not fixed file tables). Until a real fix is
done, CONNECT tries to ignore indexing and if it cannot do it
abort the command with an error message.
- Add tests on partitioning
added:
storage/connect/mysql-test/connect/r/part_file.result
storage/connect/mysql-test/connect/r/part_table.result
storage/connect/mysql-test/connect/t/part_file.test
storage/connect/mysql-test/connect/t/part_table.test
- Temporary fix
modified:
sql/sql_partition.cc
- Add partition support
modified:
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
storage/connect/reldef.cpp
storage/connect/reldef.h
storage/connect/tabdos.cpp
- Add functions ha_connect::IsUnique and ha_connect::CheckColumnList
modified:
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
- Prevent updating a partition table column that is part of
the partition function (outward tables only)
modified:
storage/connect/ha_connect.cc
- Support INSERT/UPDATE/DELETE for PROXY tables
modified:
storage/connect/tabutil.cpp
- Handle the bug on updating rows via indexing. Waiting for a real fix,
Don't use indexing when possible else raise an error and abort.
modified:
storage/connect/ha_connect.cc
- dbuserp->UseTemp set to TMP_AUTO
modified:
storage/connect/connect.cc
- Add members nox, abort and only
modified:
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
- Add arguments nox and abort to CntCloseTable
modified:
storage/connect/connect.cc
storage/connect/connect.h
storage/connect/filamap.cpp
storage/connect/filamap.h
storage/connect/filamdbf.cpp
storage/connect/filamdbf.h
storage/connect/filamfix.cpp
storage/connect/filamfix.h
storage/connect/filamtxt.cpp
storage/connect/filamtxt.h
storage/connect/filamvct.cpp
storage/connect/filamvct.h
storage/connect/filamzip.cpp
storage/connect/filamzip.h
storage/connect/ha_connect.cc
- Add arguments abort to CloseTableFile and RenameTempFile
modified:
storage/connect/filamap.cpp
storage/connect/filamap.h
storage/connect/filamdbf.cpp
storage/connect/filamdbf.h
storage/connect/filamfix.cpp
storage/connect/filamfix.h
storage/connect/filamtxt.cpp
storage/connect/filamtxt.h
storage/connect/filamvct.cpp
storage/connect/filamvct.h
storage/connect/filamzip.cpp
storage/connect/filamzip.h
storage/connect/tabdos.cpp
storage/connect/tabdos.h
storage/connect/tabvct.cpp
storage/connect/xtable.h
- Fix info->records when file does not exists
modified:
storage/connect/connect.cc
- Close XML table when opened for info
modified:
storage/connect/connect.cc
- Add function VCTFAM::GetFileLength
modified:
storage/connect/filamvct.cpp
storage/connect/filamvct.h
- Column option DISTRIB -> ENUM
modified:
storage/connect/ha_connect.cc
- Options connect, query_string and partname allways available
modified:
storage/connect/ha_connect.cc
- Add function MYSQLC::GetTableSize
modified:
storage/connect/myconn.cpp
storage/connect/myconn.h
- Add new special columns (PARTNAME, FNAME, FPATH, FTYPE and FDISK)
modified:
storage/connect/colblk.cpp
storage/connect/colblk.h
storage/connect/plgdbsem.h
storage/connect/table.cpp
- Add function ExtractFromPath
modified:
storage/connect/colblk.cpp
storage/connect/plgdbsem.h
storage/connect/plgdbutl.cpp
- Enhance Cardinality for some table types
modified:
storage/connect/tabdos.cpp
storage/connect/tabmysql.cpp
storage/connect/tabmysql.h
storage/connect/tabodbc.cpp
storage/connect/tabodbc.h
storage/connect/tabsys.cpp
storage/connect/tabsys.h
storage/connect/xindex.cpp
storage/connect/xindex.h
storage/connect/xtable.h
- Add test on special column
modified:
storage/connect/tabfmt.cpp
- Add new files (added for block indexing)
modified:
storage/connect/CMakeLists.txt
------------------------------------------------------------
revno: 4007 [merge]
committer: Olivier Bertrand <bertrandop@gmail.com>
branch nick: 10.0-connect
timestamp: Sat 2014-05-31 12:31:26 +0200
message:
- Begin adding support of partition tables
modified:
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
storage/connect/reldef.cpp
- Add INSERT/UPDATE support to PROXY tables
modified:
storage/connect/tabutil.cpp
storage/connect/tabutil.h
- Take care of SPECIAL columns
modified:
storage/connect/filamdbf.cpp
storage/connect/reldef.h
storage/connect/tabfmt.cpp
-Typo and misc
modified:
storage/connect/odbconn.cpp
storage/connect/tabfix.cpp
storage/connect/xindex.cpp
------------------------------------------------------------
revno: 4006
committer: Olivier Bertrand <bertrandop@gmail.com>
branch nick: 10.0-connect
timestamp: Sat 2014-05-10 12:21:08 +0200
message:
- FIX some MAP and XMAP errors (such as mapped indexes not closed)
Do not put version in XML files header
Remove HTON_NO_PARTITION for testing
Fix a wrong return (instead of DBUG_RETURN) in index_init
Plus a few typos
modified:
storage/connect/connect.cc
storage/connect/filter.cpp
storage/connect/ha_connect.cc
storage/connect/maputil.cpp
storage/connect/mysql-test/connect/r/alter_xml.result
storage/connect/mysql-test/connect/r/xml.result
storage/connect/table.cpp
storage/connect/tabxml.cpp
storage/connect/xindex.cpp
storage/connect/xindex.h
storage/connect/xtable.h
------------------------------------------------------------
revno: 4005
committer: Olivier Bertrand <bertrandop@gmail.com>
branch nick: 10.0-connect
timestamp: Fri 2014-05-02 15:55:45 +0200
message:
- Adding fetched columns to Dynamic index key (unique only)
Fix two bugs concerning added KXYCOL's:
1 - Not set during reading
2 - Val_K not set in FastFind
modified:
storage/connect/connect.cc
storage/connect/filamtxt.h
storage/connect/tabdos.cpp
storage/connect/tabfix.cpp
storage/connect/table.cpp
storage/connect/valblk.h
storage/connect/xindex.cpp
storage/connect/xindex.h
storage/connect/xtable.h
------------------------------------------------------------
revno: 4003
committer: Olivier Bertrand <bertrandop@gmail.com>
branch nick: 10.0-connect
timestamp: Wed 2014-04-30 10:48:29 +0200
message:
- Implementation of adding selected columns to dynamic indexes.
modified:
storage/connect/connect.cc
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
storage/connect/tabdos.cpp
storage/connect/tabdos.h
storage/connect/tabvct.cpp
storage/connect/tabvct.h
storage/connect/xindex.cpp
storage/connect/xindex.h
------------------------------------------------------------
revno: 4001
committer: Olivier Bertrand <bertrandop@gmail.com>
branch nick: 10.0-connect
timestamp: Sat 2014-04-26 00:17:26 +0200
message:
- Implement dynamic indexing
modified:
storage/connect/connect.cc
storage/connect/filter.cpp
storage/connect/filter.h
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
storage/connect/tabdos.cpp
storage/connect/tabdos.h
storage/connect/table.cpp
storage/connect/xindex.cpp
storage/connect/xindex.h
storage/connect/xtable.h
------------------------------------------------------------
revno: 3995
committer: Olivier Bertrand <bertrandop@gmail.com>
branch nick: 10.0-connect
timestamp: Sun 2014-03-23 18:49:19 +0100
message:
- Work in progress
modified:
storage/connect/filter.h
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
storage/connect/mysql-test/connect/r/alter.result
storage/connect/mysql-test/connect/r/xml.result
------------------------------------------------------------
revno: 3991
committer: Olivier Bertrand <bertrandop@gmail.com>
branch nick: 10.0-connect
timestamp: Mon 2014-03-10 18:59:36 +0100
message:
- Adding files needed for block indexing
added:
storage/connect/array.cpp
storage/connect/array.h
storage/connect/blkfil.cpp
storage/connect/blkfil.h
storage/connect/filter.cpp
storage/connect/filter.h
========================================================================
This commit of the main branch adds:
- A change needed to have the engine function check_if_supported_inplace_alter
called for partition tables (was done manually in the sub-branch) by adding
the preparser define: PARTITION_SUPPORTS_INPLACE_ALTER
modified:
sql/CMakeLists.txt
- A fix concerning the FileExists function. It was needed to force the function
table_flags to return the same flags for all partitions. This is tested by
the partition engine and raises an error if flags are not equal.
The way file name, table name and connection string are retrieved has been
modified to cope with it.
modified:
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
storage/connect/reldef.cpp
- A few typos, such as the version string.
modified:
storage/connect/ha_connect.cc
- Updating some test result files because some warnings are no more raised.
modified:
storage/connect/mysql-test/connect/r/occur.result
storage/connect/mysql-test/connect/r/part_file.result
storage/connect/mysql-test/connect/r/pivot.result
Reason for the problem was that the hash of changed files in the key cache was too small (was 128). Fixed by making the hash size larger and changeable.
- Introduced key-cache-file-hash-size (default 512) for MyISAM and aria_pagecache_file_hash_size (default 512) for Aria.
- Added new status variable "Feature_delay_key_write" which counts number of tables opened that are using delay_key_write
mysql-test/r/features.result:
Added test of Feature_delay_key_write
mysql-test/r/key_cache.result:
Updated tests as the number of blocks has changed
mysql-test/r/mysqld--help.result:
Updated result
mysql-test/suite/maria/maria3.result:
Updated result
mysql-test/suite/sys_vars/r/key_cache_file_hash_size_basic.result:
Test new variable
mysql-test/suite/sys_vars/t/aria_pagecache_file_hash_size_basic.test:
Test new variable
mysql-test/suite/sys_vars/t/key_cache_file_hash_size_basic.test:
Test new variable
mysql-test/t/features.test:
Added test of Feature_delay_key_write
mysql-test/t/key_cache.test:
Updated tests as the number of blocks has changed
mysys/mf_keycache.c:
Made CHANGED_BLOCKS_HASH dynamic
sql/handler.cc:
Updated call to init_key_cache()
sql/mysqld.cc:
Added "Feature_delay_key_write"
Added support for key-cache-file-hash-size
sql/mysqld.h:
Added support for key-cache-file-hash-size
sql/sql_class.h:
Added feature_files_opened_with_delayed_keys
sql/sys_vars.cc:
Added key_cache_file_hash_size
storage/maria/ha_maria.cc:
Added pagecache_file_hash_size
Added counting of files with delay_key_write
storage/maria/ma_checkpoint.c:
Fixed compiler warning
storage/maria/ma_pagecache.c:
Made PAGECACHE_CHANGED_BLOCKS_HASH into a variable
storage/maria/ma_pagecache.h:
Made PAGECACHE_CHANGED_BLOCKS_HASH into a variable
storage/maria/ma_rt_test.c:
Updated parameters for init_pagecache()
storage/maria/ma_test1.c:
Updated parameters for init_pagecache()
storage/maria/ma_test2.c:
Updated parameters for init_pagecache()
storage/maria/ma_test3.c:
Updated parameters for init_pagecache()
storage/maria/maria_chk.c:
Updated parameters for init_pagecache()
storage/maria/maria_ftdump.c:
Updated parameters for init_pagecache()
storage/maria/maria_pack.c:
Updated parameters for init_pagecache()
storage/maria/maria_read_log.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_pagecache_consist.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_pagecache_rwconsist.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_pagecache_rwconsist2.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_pagecache_single.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_test_loghandler-t.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_test_loghandler_first_lsn-t.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_test_loghandler_max_lsn-t.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_test_loghandler_multigroup-t.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_test_loghandler_multithread-t.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_test_loghandler_noflush-t.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_test_loghandler_nologs-t.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_test_loghandler_pagecache-t.c:
Updated parameters for init_pagecache()
storage/maria/unittest/ma_test_loghandler_purge-t.c:
Updated parameters for init_pagecache()
storage/myisam/ha_myisam.cc:
Added counting of files with delay_key_write
storage/myisam/mi_check.c:
Updated call to init_key_cache()
storage/myisam/mi_test1.c:
Updated call to init_key_cache()
storage/myisam/mi_test2.c:
Updated call to init_key_cache()
storage/myisam/mi_test3.c:
Updated call to init_key_cache()
storage/myisam/mi_test_all.sh:
Fixed broken test
storage/myisam/myisam_ftdump.c:
Updated call to init_key_cache()
storage/myisam/myisamchk.c:
Updated call to init_key_cache()
storage/myisam/myisamlog.c:
Updated call to init_key_cache()
chain
InnoDB mutex_exit() function calls __sync_test_and_set() to release
the lock. According to manual this function is supposed to create
"acquire" memory barrier whereas in fact we need "release" memory
barrier at mutex_exit().
The problem isn't repeatable with gcc because it creates
"acquire-release" memory barrier for __sync_test_and_set().
ATC creates just "acquire" barrier.
Fixed by creating proper barrier at mutex_exit() by using
__sync_lock_release() instead of __sync_test_and_set().
1) Support of partitioning by connect. A table can be partitioned
by files, this is an enhanced MULTIPLE table. It can be also
partitioned by sub-tables like TBL and this enables table sharding.
2) Handling a CONNECT bug that causes in some cases extraneous rows
to remain in the table after an UPDATE or DELETE when the command
uses indexing (for not fixed file tables). Until a real fix is
done, CONNECT tries to ignore indexing and if it cannot do it
abort the command with an error message.
- Add tests on partitioning
added:
storage/connect/mysql-test/connect/r/part_file.result
storage/connect/mysql-test/connect/r/part_table.result
storage/connect/mysql-test/connect/t/part_file.test
storage/connect/mysql-test/connect/t/part_table.test
- Temporary fix
modified:
sql/sql_partition.cc
- Add partition support
modified:
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
storage/connect/reldef.cpp
storage/connect/reldef.h
storage/connect/tabdos.cpp
- Add functions ha_connect::IsUnique and ha_connect::CheckColumnList
modified:
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
- Prevent updating a partition table column that is part of
the partition function (outward tables only)
modified:
storage/connect/ha_connect.cc
- Support INSERT/UPDATE/DELETE for PROXY tables
modified:
storage/connect/tabutil.cpp
- Handle the bug on updating rows via indexing. Waiting for a real fix,
Don't use indexing when possible else raise an error and abort.
modified:
storage/connect/ha_connect.cc
- dbuserp->UseTemp set to TMP_AUTO
modified:
storage/connect/connect.cc
- Add members nox, abort and only
modified:
storage/connect/ha_connect.cc
storage/connect/ha_connect.h
- Add arguments nox and abort to CntCloseTable
modified:
storage/connect/connect.cc
storage/connect/connect.h
storage/connect/filamap.cpp
storage/connect/filamap.h
storage/connect/filamdbf.cpp
storage/connect/filamdbf.h
storage/connect/filamfix.cpp
storage/connect/filamfix.h
storage/connect/filamtxt.cpp
storage/connect/filamtxt.h
storage/connect/filamvct.cpp
storage/connect/filamvct.h
storage/connect/filamzip.cpp
storage/connect/filamzip.h
storage/connect/ha_connect.cc
- Add arguments abort to CloseTableFile and RenameTempFile
modified:
storage/connect/filamap.cpp
storage/connect/filamap.h
storage/connect/filamdbf.cpp
storage/connect/filamdbf.h
storage/connect/filamfix.cpp
storage/connect/filamfix.h
storage/connect/filamtxt.cpp
storage/connect/filamtxt.h
storage/connect/filamvct.cpp
storage/connect/filamvct.h
storage/connect/filamzip.cpp
storage/connect/filamzip.h
storage/connect/tabdos.cpp
storage/connect/tabdos.h
storage/connect/tabvct.cpp
storage/connect/xtable.h
- Fix info->records when file does not exists
modified:
storage/connect/connect.cc
- Close XML table when opened for info
modified:
storage/connect/connect.cc
- Add function VCTFAM::GetFileLength
modified:
storage/connect/filamvct.cpp
storage/connect/filamvct.h
- Column option DISTRIB -> ENUM
modified:
storage/connect/ha_connect.cc
- Options connect, query_string and partname allways available
modified:
storage/connect/ha_connect.cc
- Add function MYSQLC::GetTableSize
modified:
storage/connect/myconn.cpp
storage/connect/myconn.h
- Add new special columns (PARTNAME, FNAME, FPATH, FTYPE and FDISK)
modified:
storage/connect/colblk.cpp
storage/connect/colblk.h
storage/connect/plgdbsem.h
storage/connect/table.cpp
- Add function ExtractFromPath
modified:
storage/connect/colblk.cpp
storage/connect/plgdbsem.h
storage/connect/plgdbutl.cpp
- Enhance Cardinality for some table types
modified:
storage/connect/tabdos.cpp
storage/connect/tabmysql.cpp
storage/connect/tabmysql.h
storage/connect/tabodbc.cpp
storage/connect/tabodbc.h
storage/connect/tabsys.cpp
storage/connect/tabsys.h
storage/connect/xindex.cpp
storage/connect/xindex.h
storage/connect/xtable.h
- Add test on special column
modified:
storage/connect/tabfmt.cpp
- Add new files (added for block indexing)
modified:
storage/connect/CMakeLists.txt
Merge the patches into MariaDB 10.0 main.
With this patch, parallel replication will now automatically retry a
transaction that fails due to deadlock or other temporary error, same as
single-threaded replication.
We catch deadlocks with InnoDB transactions due to enforced commit order. If
T1 must commit before T2 in parallel replication and T1 ends up waiting for T2
inside InnoDB, we kill T2 and retry it later to resolve the deadlock
automatically.
After-review changes.
For this patch in 10.0, we do not introduce a new public storage engine API,
we just fix the InnoDB/XtraDB issues. In 10.1, we will make a better public
API that can be used for all storage engines (MDEV-6429).
Eliminate the background thread that did deadlock kills asynchroneously.
Instead, we ensure that the InnoDB/XtraDB code can handle doing the kill from
inside the deadlock detection code (when thd_report_wait_for() needs to kill a
later thread to resolve a deadlock).
(We preserve the part of the original patch that introduces dedicated mutex
and condition for the slave init thread, to remove the abuse of
LOCK_thread_count for start/stop synchronisation of the slave init thread).
Analysis: For some reason table stats for a table pointed from a index
is not initialized. Added additional warning output on this situation
and table stats initialization. This is better than asserting.
than with InnoDB plugin
Fix: os0file.h in XtraDB had OS_AIO_N_PENDING_IOS_PER_THREAD 256
when on InnoDB it is OS_AIO_N_PENDING_IOS_PER_THREAD 32. Changed
XtraDB also to use 32.
WITH CERTAIN MAX_HEAP_TABLE_SIZE VALUES
Description:
When the system variable 'max_heap_table_size'
is set to 20GB, the server crashes on creation of a
temporary tables or tables using MEMORY storage engine.
Analysis:
The variable 'max_record' determines the amount heap
allocated for the records of the table. This value
is determined using the 'max_heap_table_size' variable.
'records_in_block' in turn uses the max_records to
determine the number of records per block.
When the 'max_heap_table_size' is set to 20GB, then
the 'records_in_block' is calculated to a value of
2^28.
The size of the block determined by multiplying the
'records_in_block' and 'recbuffer' results in overflow
and hence the value becomes zero. As a result, zero bytes
of the heap is allocated for the table. This will
result in a server crash when the table is accessed.
Fix:
The variables 'records_in_block' and 'recbuffer' are
typecasted to 'unsigned long' while calculating the
size of the block.
than with InnoDB plugin
Fix: os0file.h in XtraDB had OS_AIO_N_PENDING_IOS_PER_THREAD 256
when on InnoDB it is OS_AIO_N_PENDING_IOS_PER_THREAD 32. Changed
XtraDB also to use 32.
Analysis: Based on crashed the buffer pool instance identifier is
not correct on block to be freed. Add LRU list mutex holding
on functions calling free and add additional safety checks.
replication causing replication to fail.
Remove the temporary fix for MDEV-5914, which used READ COMMITTED for parallel
replication worker threads. Replace it with a better, more selective solution.
The issue is with certain edge cases of InnoDB gap locks, for example between
INSERT and ranged DELETE. It is possible for the gap lock set by the DELETE to
block the INSERT, if the DELETE runs first, while the record lock set by
INSERT does not block the DELETE, if the INSERT runs first. This can cause a
conflict between the two in parallel replication on the slave even though they
ran without conflicts on the master.
With this patch, InnoDB will ask the server layer about the two involved
transactions before blocking on a gap lock. If the server layer tells InnoDB
that the transactions are already fixed wrt. commit order, as they are in
parallel replication, InnoDB will ignore the gap lock and allow the two
transactions to proceed in parallel, avoiding the conflict.
Improve the fix for MDEV-6020. When InnoDB itself detects a deadlock, it now
asks the server layer for any preferences about which transaction to roll
back. In case of parallel replication with two transactions T1 and T2 fixed to
commit T1 before T2, the server layer will ask InnoDB to roll back T2 as the
deadlock victim, not T1. This helps in some cases to avoid excessive deadlock
rollback, as T2 will in any case need to wait for T1 to complete before it can
itself commit.
Also some misc. fixes found during development and testing:
- Remove thd_rpl_is_parallel(), it is not used or needed.
- Use KILL_CONNECTION instead of KILL_QUERY when a parallel replication
worker thread is killed to resolve a deadlock with fixed commit
ordering. There are some cases, eg. in sql/sql_parse.cc, where a KILL_QUERY
can be ignored if the query otherwise completed successfully, and this
could cause the deadlock kill to be lost, so that the deadlock was not
correctly resolved.
- Fix random test failure due to missing wait_for_binlog_checkpoint.inc.
- Make sure that deadlock or other temporary errors during parallel
replication are not printed to the the error log; there were some places
around the replication code with extra error logging. These conditions can
occur occasionally and are handled automatically without breaking
replication, so they should not pollute the error log.
- Fix handling of rgi->gtid_sub_id. We need to be able to access this also at
the end of a transaction, to be able to detect and resolve deadlocks due to
commit ordering. But this value was also used as a flag to mark whether
record_gtid() had been called, by being set to zero, losing the value. Now,
introduce a separate flag rgi->gtid_pending, so rgi->gtid_sub_id remains
valid for the entire duration of the transaction.
- Fix one place where the code to handle ignored errors called reset_killed()
unconditionally, even if no error was caught that should be ignored. This
could cause loss of a deadlock kill signal, breaking deadlock detection and
resolution.
- Fix a couple of missing mysql_reset_thd_for_next_command(). This could
cause a prior error condition to remain for the next event executed,
causing assertions about errors already being set and possibly giving
incorrect error handling for following event executions.
- Fix code that cleared thd->rgi_slave in the parallel replication worker
threads after each event execution; this caused the deadlock detection and
handling code to not be able to correctly process the associated
transactions as belonging to replication worker threads.
- Remove useless error code in slave_background_kill_request().
- Fix bug where wfc->wakeup_error was not cleared at
wait_for_commit::unregister_wait_for_prior_commit(). This could cause the
error condition to wrongly propagate to a later wait_for_prior_commit(),
causing spurious ER_PRIOR_COMMIT_FAILED errors.
- Do not put the binlog background thread into the processlist. It causes
too many result differences in mtr, but also it probably is not useful
for users to pollute the process list with a system thread that does not
really perform any user-visible tasks...
SLOW/CRASHES SEMAPHORE
Problem:
There are 2 lakh tables - fk_000001, fk_000002 ... fk_200000. All of them
are related to the same parent_table through a foreign key constraint.
When the parent_table is loaded into the dictionary cache, all the child table
will also be loaded. This is taking lot of time. Since this operation happens
when the dictionary latch is taken, the scenario leads to "long semaphore wait"
situation and the server gets killed.
Analysis:
A simple performance analysis showed that the slowness is because of the
dict_foreign_find() function. It does a linear search on two linked list
table->foreign_list and table->referenced_list, looking for a particular
foreign key object based on foreign->id as the key. This is called two
times for each foreign key object.
Solution:
Introduce a rb tree in table->foreign_rbt and table->referenced_rbt, which
are some sort of index on table->foreign_list and table->referenced_list
respectively, using foreign->id as the key. These rbt structures will be
solely used by dict_foreign_find().
rb#5599 approved by Vasil
