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.
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.
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...
Due to how gap locks work, two transactions could group commit together on the
master, but get lock conflicts and then deadlock due to different thread
scheduling order on slave.
For now, remove these deadlocks by running the parallel slave in READ
COMMITTED mode. And let InnoDB/XtraDB allow statement-based binlogging for the
parallel slave in READ COMMITTED.
We are also investigating a different solution long-term, which is based on
relaxing the gap locks only between the transactions running in parallel for
one slave, but not against possibly external transactions.
Syntax. Server support. Test cases.
InnoDB bugfixes:
* don't mess around with system sprintf's, always use my_error() for errors.
* don't use InnoDB internal error codes where OS error codes are expected.
* don't say "file not found", when it was.
Update InnoDB to 5.6.14
Apply MySQL-5.6 hack for MySQL Bug#16434374
Move Aria-only HA_RTREE_INDEX from my_base.h to maria_def.h (breaks an assert in InnoDB)
Fix InnoDB memory leak
support ha_innodb.so as a dynamic plugin.
* remove obsolete *,innodb_plugin.rdiff files
* s/--plugin-load=/--plugin-load-add=/
* MYSQL_PLUGIN_IMPORT glob_hostname[]
* use my_error instead of push_warning_printf(ER_DEFAULT)
* don't use tdc_size and tc_size in a module
update test cases (XtraDB is 5.6.14, InnoDB is 5.6.10)
* copy new tests over
* disable some tests for (old) InnoDB
* delete XtraDB tests that no longer apply
small compatibility changes:
* s/HTON_EXTENDED_KEYS/HTON_SUPPORTS_EXTENDED_KEYS/
* revert unnecessary InnoDB changes to make it a bit closer to the upstream
fix XtraDB to compile on Windows (both as a static and a dynamic plugin)
disable XtraDB on Windows (deadlocks) and where no atomic ops are available (e.g. CentOS 5)
storage/innobase/handler/ha_innodb.cc:
revert few unnecessary changes to make it a bit closer to the original InnoDB
storage/innobase/include/univ.i:
correct the version to match what it was merged from
restore old innodb get_innobase_type_from_mysql_type() function,
record all mysql_type->innodb_type mapping
(as generated by mysql-5.6).
add safety code to disable online alter when internal types don't match
storage/innobase/dict/dict0stats.cc:
revert to 5.6 state
The merge is still missing a few hunks related to temporary tables and
InnoDB log file size. The associated code did not seem to exist in
10.0, so the merge of that needs more work. Until this is fixed, there
are a number of test failures as a result.
For compatibility purposes let InnoDB use DATA_INT for MYSQL_TYPE_ENUM and MYSQL_TYPE_SET.
Silence the warning for these types and let the index translation table to be built anyway.
Test case by Jeremy Cole.
Add an error code to the wait_for_commit facility.
Now, when a transaction fails, it can signal the error to
any subsequent transaction that is waiting for it to commit.
The waiting transactions then receive the error code back from
wait_for_prior_commit() and can handle the error appropriately.
Also fix one race that could cause crash if @@slave_parallel_threads
were changed several times quickly in succession.
Following variables do not require LOCK_open protection anymore:
- table_def_cache (renamed to tdc_hash) is protected by rw-lock
LOCK_tdc_hash;
- table_def_shutdown_in_progress doesn't need LOCK_open protection;
- last_table_id use atomics;
- TABLE_SHARE::ref_count (renamed to TABLE_SHARE::tdc.ref_count)
is protected by TABLE_SHARE::tdc.LOCK_table_share;
- TABLE_SHARE::next, ::prev (renamed to tdc.next and tdc.prev),
oldest_unused_share, end_of_unused_share are protected by
LOCK_unused_shares;
- TABLE_SHARE::m_flush_tickets (renamed to tdc.m_flush_tickets)
is protected by TABLE_SHARE::tdc.LOCK_table_share;
- refresh_version (renamed to tdc_version) use atomics.
remove unused HA_CAN_EXPORT
(with the merge of "FLUSH TABLE <table_list> FOR EXPORT",
this functionality should be added back, but differently -
no table flag, handler::extra() returns an error instead)
Problem:
When the user specified foreign key name contains "_ibfk_", InnoDB wrongly
tries to rename it.
Solution:
When a table is renamed, all its associated foreign keys will also be renamed,
only if the foreign key names are automatically generated. If the foreign key
names are given by the user, even if it has _ibfk_ in it, it must not be
renamed.
rb#2935 approved by Jimmy, Krunal and Satya
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
SERIALIZABLE
Problem:
The documentation claims that WITH CONSISTENT SNAPSHOT will work for both
REPEATABLE READ and SERIALIZABLE isolation levels. But it will work only
for REPEATABLE READ isolation level. Also, the clause WITH CONSISTENT
SNAPSHOT is silently ignored when it is not applicable to the given isolation
level.
Solution:
Generate a warning when the clause WITH CONSISTENT SNAPSHOT is ignored.
rb#2797 approved by Kevin.
Note: Support team wanted to push this to 5.5+.
Merged all ddl_logging code.
Merged sql_partition.cc
innodb_mysql_lock2.test and partition_cache.test now works.
Changed interface to strconvert() to make it easier to use with not \0 terminated strings.
sql/sql_partition.cc:
Full merge with 5.6
sql/sql_table.cc:
Merged all ddl_logging code
sql/strfunc.cc:
Added from_length argument to strconvert() to make it possible to use without end terminated strings.
sql/strfunc.h:
Added from_length argument to strconvert() to make it possible to use without end terminated strings.
mysql-test/r/mdl_sync.result:
Full merge with 5.6
mysql-test/t/mdl_sync.test:
Full merge with 5.6
sql/debug_sync.cc:
Full merge with 5.6
sql/debug_sync.h:
Full merge with 5.6
sql/mdl.cc:
Full merge with 5.6
sql/sql_base.cc:
Removed code not in 5.6 anymore
Implement facility for the commit in one thread to wait for the commit of
another to complete first. The wait is done in a way that does not hinder
that a waiter and a waitee can group commit together with a single fsync()
in both binlog and InnoDB. The wait is done efficiently with respect to
locking.
The patch was originally made to support TaoBao parallel replication with
in-order commit; now it will be adapted to also be used for parallel
replication of group-committed transactions.
A waiter THD registers itself with a prior waitee THD. The waiter will then
complete its commit at the earliest in the same group commit of the waitee
(when using binlog). The wait can also be done explicitly by the waitee.
Analysis
--------
The pthread_mutex commit_threads_m was initiliazed but never
used.
Fix
---
Removing the commit_threads_m mutex from the code base.
[ Approved by Marko rb#2475]
DDL AND I_S QUERIES
Skip partially created indexes (ones whose name starts with TEMP_INDEX_PREFIX)
from stats gathering.
Because InnoDB reports HA_INPLACE_ADD_INDEX_NO_WRITE to MySQL, the latter
allows parallel execution of ha_innobase::add_index() and ha_innobase::info().
Reviewed by: Inaam (rb:2613)
IF IT HAS A WRONG COUNT
If CHECK TABLE finds that a secondary index contains the wrong
number of entries, it used to report an error but not mark the
index as corrupt. The error means that the index should be rebuilt,
which can be done with ALTER TABLE DROP INDEX and ALTER TABLE ADD
INDEX. But just in case the DBA does not pay any attention to the
output of CHECK TABLE, the secondary index should be marked as
corrupted so that it is not used again.
Approved by Inaam in RB:2607
AFTER A ROW IS READ
Approved by: Sunny Bains rb://2425
Don't release concurrency tickets when asked to release
btr_search_latch. This is a 5.5 only bug. It is already
fixed in 5.6 upwards.
