Some DML operations on tables having unique secondary keys cause scanning
in the secondary index, for instance to find potential unique key violations
in the seconday index. This scanning may involve GAP locking in the index.
As this locking happens also when applying replication events in high priority
applier threads, there is a probabality for lock conflicts between two wsrep
high priority threads.
This PR avoids lock conflicts of high priority wsrep threads, which do
secondary index scanning e.g. for duplicate key detection.
The actual fix is the patch in sql_class.cc:thd_need_ordering_with(), where
we allow relaxed GAP locking protocol between wsrep high priority threads.
wsrep high priority threads (replication appliers, replayers and TOI processors)
are ordered by the replication provider, and they will not need serializability
support gained by secondary index GAP locks.
PR contains also a mtr test, which exercises a scenario where two replication
applier threads have a false positive conflict in GAP of unique secondary index.
The conflicting local committing transaction has to replay, and the test verifies
also that the replaying phase will not conflict with the latter repllication applier.
Commit also contains new test scenario for galera.galera_UK_conflict.test,
where replayer starts applying after a slave applier thread, with later seqno,
has advanced to commit phase. The applier and replayer have false positive GAP
lock conflict on secondary unique index, and replayer should ignore this.
This test scenario caused crash with earlier version in this PR, and to fix this,
the secondary index uniquenes checking has been relaxed even further.
Now innodb trx_t structure has new member: bool wsrep_UK_scan, which is set to
true, when high priority thread is performing unique secondary index scanning.
The member trx_t::wsrep_UK_scan is defined inside WITH_WSREP directive, to make
it possible to prepare a MariaDB build where this additional trx_t member is
not present and is not used in the code base. trx->wsrep_UK_scan is set to true
only for the duration of function call for: lock_rec_lock() trx->wsrep_UK_scan
is used only in lock_rec_has_to_wait() function to relax the need to wait if
wsrep_UK_scan is set and conflicting transaction is also high priority.
Reviewed-by: Jan Lindström <jan.lindstrom@mariadb.com>
The reason for the failure is that
thd->mdl_context.release_transactional_locks()
was called after commit & rollback even in cases where the current
transaction is still active.
For 10.2, 10.3 and 10.4 the fix is simple:
- Replace all calls to thd->mdl_context.release_transactional_locks() with
thd->release_transactional_locks(). The thd function will only call
the mdl_context function if there are no active transactional locks.
In 10.6 we will better fix where we will change the return value for
some trans_xxx() functions to indicate if transaction did close the
transaction or not. This will avoid the need of the indirect call.
Other things:
- trans_xa_commit() and trans_xa_rollback() will automatically
call release_transactional_locks() if the transaction is closed.
- We can't do that for the other functions as the caller of many of these
are doing additional work (like close_thread_tables) before calling
release_transactional_locks().
- Added missing abort_result_set() and missing DBUG_RETURN in
select_create::send_eof()
- Fixed wrong indentation in injector::transaction::commit()
Other things, mainly to get
create_mysqld_error_find_printf_error tool to work:
- Added protection to not include mysqld_error.h twice
- Include "unireg.h" instead of "mysqld_error.h" in server
- Added protection if ER_XX messages are already defined
- Removed wrong calls to my_error(ER_OUTOFMEMORY) as
my_malloc() and my_alloc will do this automatically
- Added missing %s to ER_DUP_QUERY_NAME
- Removed old and wrong calls to my_strerror() when using
MY_ERROR_ON_RENAME (wrong merge)
- Fixed deadlock error message from Galera. Before the extra
information given to ER_LOCK_DEADLOCK was missing because
ER_LOCK_DEADLOCK doesn't provide any extra information.
I kept #ifdef mysqld_error_find_printf_error_used in sql_acl.h
to make it easy to do this kind of check again in the future
Relaxed assertion (in MySQL it was removed).
For "LOCK TABLES t1 WRITE CONCURRENT, t1 READ" upgrade lock to weakest
existing suitable lock, which is MDL_SHARED_NO_READ_WRITE.
Implementation of MDEV-7660 introduced unwanted incompatible change:
modifications under LOCK TABLES with autocommit enabled are rolled back on
disconnect. Previously everything was committed, because LOCK TABLES didn't
adjust autocommit setting.
This patch restores original behavior by reverting some changes done in
MDEV-7660:
- sql/sql_parse.cc: do not reset autocommit on LOCK TABLES
- sql/sql_base.cc: do not set autocommit on UNLOCK TABLES
- test cases: main.lock_tables_lost_commit, main.partition_explicit_prune,
rpl.rpl_switch_stm_row_mixed, tokudb.nested_txn_implicit_commit,
tokudb_bugs.db806
But it makes InnoDB tables under LOCK TABLES ... READ [LOCAL] not protected
against DML. To restore protection some changes from WL#6671 were merged,
specifically MDL_SHARED_READ_ONLY and test cases.
WL#6671 merge highlights:
- Not all tests merged.
- In MySQL LOCK TABLES ... READ acquires MDL_SHARED_READ_ONLY for all engines,
in MariaDB MDL_SHARED_READ is always acquired first and then upgraded to
MDL_SHARED_READ_ONLY for InnoDB only.
- The above allows us to omit MDL_SHARED_WRITE_LOW_PRIO implementation in
MariaDB, which is rather useless with InnoDB. In MySQL it is needed to
preserve locking behavior between low priority writes and LOCK TABLES ... READ
for non-InnoDB engines (covered by sys_vars.sql_low_priority_updates_func).
- Omitted HA_NO_READ_LOCAL_LOCK, we rely on lock_count() instead.
- Omitted "piglets": in MariaDB stream of DML against InnoDB table may lead to
concurrent LOCK TABLES ... READ starvation.
- HANDLER ... OPEN acquires MDL_SHARED_READ instead of MDL_SHARED in MariaDB.
- Omitted SNRW->X MDL lock upgrade for IMPORT/DISCARD TABLESPAECE under LOCK
TABLES.
- Omitted strong locks for views, triggers and SP under LOCK TABLES.
- Omitted IX schema lock for LOCK TABLES READ.
- Omitted deadlock weight juggling for LOCK TABLES.
Full WL#6671 merge status:
- innodb.innodb-lock: fully merged
- main.alter_table: not merged due to different HANDLER solution
- main.debug_sync: fully merged
- main.handler_innodb: not merged due to different HANDLER solution
- main.handler_myisam: not merged due to different HANDLER solution
- main.innodb_mysql_lock: fully merged
- main.insert_notembedded: fully merged
- main.lock: not merged (due to no strong locks for views)
- main.lock_multi: not merged
- main.lock_sync: fully merged (partially in MDEV-7660)
- main.mdl_sync: not merged
- main.partition_debug_sync: not merged due to different HANDLER solution
- main.status: fully merged
- main.view: fully merged
- perfschema.mdl_func: not merged (no such test in MariaDB)
- perfschema.table_aggregate_global_2u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_global_2u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_global_4u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_global_4u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_hist_2u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_hist_2u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_hist_4u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_hist_4u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_thread_2u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_thread_2u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_thread_4u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_aggregate_thread_4u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_global_2u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_global_2u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_global_4u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_global_4u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_hist_2u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_hist_2u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_hist_4u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_hist_4u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_thread_2u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_thread_2u_3t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_thread_4u_2t: not merged (didn't fail in MariaDB)
- perfschema.table_lock_aggregate_thread_4u_3t: not merged (didn't fail in MariaDB)
- sys_vars.sql_low_priority_updates_func: not merged
- include/thr_rwlock.h: not merged, rw_pr_lock_assert_write_owner and
rw_pr_lock_assert_not_write_owner are macros in MariaDB
- sql/handler.h: not merged (HA_NO_READ_LOCAL_LOCK)
- sql/mdl.cc: partially merged (MDL_SHARED_READ_ONLY only)
- sql/mdl.h: partially merged (MDL_SHARED_READ_ONLY only)
- sql/lock.cc: fully merged
- sql/sp_head.cc: not merged
- sql/sp_head.h: not merged
- sql/sql_base.cc: partially merged (MDL_SHARED_READ_ONLY only)
- sql/sql_base.h: not merged
- sql/sql_class.cc: fully merged
- sql/sql_class.h: fully merged
- sql/sql_handler.cc: merged partially (different solution in MariaDB)
- sql/sql_parse.cc: partially merged, mostly omitted low priority write part
- sql/sql_reload.cc: not merged comment change
- sql/sql_table.cc: not merged SNRW->X upgrade for IMPORT/DISCARD TABLESPACE
- sql/sql_view.cc: not merged
- sql/sql_yacc.yy: not merged (MDL_SHARED_WRITE_LOW_PRIO, MDL_SHARED_READ_ONLY)
- sql/table.cc: not merged (MDL_SHARED_WRITE_LOW_PRIO)
- sql/table.h: not merged (MDL_SHARED_WRITE_LOW_PRIO)
- sql/trigger.cc: not merged
- storage/innobase/handler/ha_innodb.cc: merged store_lock()/lock_count()
changes (in MDEV-7660), didn't merge HA_NO_READ_LOCAL_LOCK
- storage/innobase/handler/ha_innodb.h: fully merged in MDEV-7660
- storage/myisammrg/ha_myisammrg.cc: not merged comment change
- storage/perfschema/table_helper.cc: not merged (no MDL support in MariaDB PFS)
- unittest/gunit/mdl-t.cc: not merged
- unittest/gunit/mdl_sync-t.cc: not merged
MariaDB specific changes:
- handler.heap: different HANDLER solution, MDEV-7660
- handler.innodb: different HANDLER solution, MDEV-7660
- handler.interface: different HANDLER solution, MDEV-7660
- handler.myisam: different HANDLER solution, MDEV-7660
- main.mdl_sync: MDEV-7660 specific changes
- main.partition_debug_sync: removed test due to different HANDLER solution,
MDEV-7660
- main.truncate_coverage: removed test due to different HANDLER solution,
MDEV-7660
- mysql-test/include/mtr_warnings.sql: additional cleanup, MDEV-7660
- mysql-test/lib/v1/mtr_report.pl: additional cleanup, MDEV-7660
- plugin/metadata_lock_info/metadata_lock_info.cc: not in MySQL
- sql/sql_handler.cc: MariaDB specific fix for mysql_ha_read(), MDEV-7660
The issue was that when running with valgrind the wait for master_pos_Wait()
was not long enough.
This patch also fixes two other failures that could affect rpl_mdev6020:
- check_if_conflicting_replication_locks() didn't properly check domains
- 'did_mark_start_commit' was after signals to other threads was sent which could
get the variable read too early.
(like DROP TABLE) has been scheduled before conflicting DDL's (like INSERT)
are commited.
What makes these bugs hard to detect is that in most cases any wrong
schduling are caught by MDL locks. It's only when there are timing issues
that the bugs (usually deadlocks) are noticed.
This patch adds a DBUG_ASSERT() that detects, in parallel replication,
if a DDL is scheduled before any depending DML'S are commited.
It does this be checking if there are any conflicting replication locks
when the DDL is about to wait for getting it's MDL lock.
I also did some minor code cleanups in sql_base.cc to make this code
similar to other related code.
metadata_lock_info plugin called MDL_context::find_ticket() to obtain lock
duration, which in turn iterates foreign thread private lists. These lists
can be updated by owner thread without protection.
Fixed by iterating threads (instead of MDL locks and tickets) and obtaining
data through APC.
Also fixed mdl_iterate_lock() to initialize iterator under prlock protection.
- avoiding the race condition, by not grabbing thd->LOCK_wsrep_thd for
accessing thd->wsrep_exec_mode. The caller is same thread and exec mode
can only be changed by self.
- Changed ER(ER_...) to ER_THD(thd, ER_...) when thd was known or if there was many calls to current_thd in the same function.
- Changed ER(ER_..) to ER_THD_OR_DEFAULT(current_thd, ER...) in some places where current_thd is not necessary defined.
- Removing calls to current_thd when we have access to thd
Part of this is optimization (not calling current_thd when not needed),
but part is bug fixing for error condition when current_thd is not defined
(For example on startup and end of mysqld)
Notable renames done as otherwise a lot of functions would have to be changed:
- In JOIN structure renamed:
examined_rows -> join_examined_rows
record_count -> join_record_count
- In Field, renamed new_field() to make_new_field()
Other things:
- Added DBUG_ASSERT(thd == tmp_thd) in Item_singlerow_subselect() just to be safe.
- Removed old 'tab' prefix in JOIN_TAB::save_explain_data() and use members directly
- Added 'thd' as argument to a few functions to avoid calling current_thd.
* Wait for aborted thd (victim) to release MDL locks
* Skip aborting an already aborted thd
* Defer setting OK status in case of CTAS
* Minor cosmetic changes
* Added a test case
Initialize abs_timeout when it is about to be used. This saves one my_hrtime()
call on hot path (when we acquire MDL lock without waiting).
When filling I_S.PROCESSLIST use THD::start_utime/THD::utime_after_query instead
of THD::start_time. This allows us to save 2 clock_gettime() calls.
Overhead change:
__clock_gettime 0.13% -> 0.11% (122 -> 76 calls per OLTP RO transaction)
my_interval_timer 0.07% -> 0.06%
my_hrtime 0.04% -> 0.01%
C++03 seem to require default constructor for const objects even though they're
fully initialized. There is defect report for this:
http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_active.html#253
Apparently some compilers (e.g. gcc) addressed this defect report, while some
(e.g. clang) did not.
Added empty default constructor to MDL_scoped_lock and MDL_object_lock classes.
Also replaced lf_hash_search_using_hash_value() with lf_hash_search(). The
latter will call mdl_key->hash_value() anyway.
Simplified MDL_object_lock and MDL_scoped_lock dichotomy so that they can be
stored in LF_HASH. This was done by moving out their differences to a
MDL_lock_strategy structure which is referenced from the MDL_lock object by
pointer.
Replaced m_ref_usage, m_ref_release, m_is_destroyed and m_version with single
m_state, which is updated atomically. Needed to get rid of MDL_map::m_mutex.
Fix MDL to report an error when a wait was killed, but preserve
the old documented behavior of GET_LOCK() where killing it is not an error.
Also remove race conditions in main.create_or_replace test
Merged lp:maria/maria-10.0-galera up to revision 3879.
Added a new functions to handler API to forcefully abort_transaction,
producing fake_trx_id, get_checkpoint and set_checkpoint for XA. These
were added for future possiblity to add more storage engines that
could use galera replication.
I_S table
mdl_iterate() helper function (which is used by the plugin to iterate mdl
locks) acquired mutexes in reverse order.
Fixed by iterating MDL locks in two stages:
1. Iterate locks hash under the protection of hash mutex, store all
lock pointers in a thread local array and increment reference counter
for the lock.
2. Iterate local array without protection of hash mutex, handle destroyed
locks.
It somewhat echoes hack in MDL_map_partition::move_from_hash_to_lock_mutex.
