Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.1(EOL), 10.2(EOL)
-
None
Description
|
Stack trace from 10.1 commit 536112dd |
10.1/sql/log_event.cc:4103: int Query_log_event::do_apply_event(rpl_group_info*, const char*, uint32): Assertion `!current_stmt_is_commit || !rgi->tables_to_lock' failed.
|
150521 19:33:11 [ERROR] mysqld got signal 6 ;
|
|
|
#6 0x00007f411ca86311 in *__GI___assert_fail (assertion=0x7f411fe19f88 "!current_stmt_is_commit || !rgi->tables_to_lock", file=<optimized out>, line=4103, function=0x7f411fe1e240 "int Query_log_event::do_apply_event(rpl_group_info*, const char*, uint32)") at assert.c:81
|
#7 0x00007f411f6e3250 in Query_log_event::do_apply_event (this=0x7f410e8b0230, rgi=0x7f410e820800, query_arg=0x7f410e885579 "COMMIT", q_len_arg=6) at 10.1/sql/log_event.cc:4103
|
#8 0x00007f411f6e2f73 in Query_log_event::do_apply_event (this=0x7f410e8b0230, rgi=0x7f410e820800) at 10.1/sql/log_event.cc:4013
|
#9 0x00007f411f32a6c9 in Log_event::apply_event (this=0x7f410e8b0230, rgi=0x7f410e820800) at 10.1/sql/log_event.h:1347
|
#10 0x00007f411f32026f in apply_event_and_update_pos (ev=0x7f410e8b0230, thd=0x7f410e849070, rgi=0x7f410e820800, rpt=0x0) at 10.1/sql/slave.cc:3274
|
#11 0x00007f411f320a86 in exec_relay_log_event (thd=0x7f410e849070, rli=0x7f410fc0ac40, serial_rgi=0x7f410e820800) at 10.1/sql/slave.cc:3577
|
#12 0x00007f411f323d68 in handle_slave_sql (arg=0x7f410fc09000) at 10.1/sql/slave.cc:4677
|
#13 0x00007f411faa61a2 in pfs_spawn_thread (arg=0x7f411c01fc70) at 10.1/storage/perfschema/pfs.cc:1860
|
#14 0x00007f411eaa1b50 in start_thread (arg=<optimized out>) at pthread_create.c:304
|
#15 0x00007f411cb3695d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:112
|
--source include/master-slave.inc
|
--source include/have_innodb.inc
|
--source include/have_binlog_format_row.inc
|
|
|
CREATE TABLE t1 (i INT) ENGINE=InnoDB; |
CREATE TABLE t2 (j INT) ENGINE=MyISAM; |
|
|
--delimiter ||
|
CREATE TRIGGER tr AFTER INSERT ON t1 FOR EACH ROW |
BEGIN |
SET @a = unknown_column_just_to_raise_an_error; |
INSERT INTO t2 VALUES (NULL) ; |
END|| |
--delimiter ;
|
|
|
--error ER_BAD_FIELD_ERROR
|
INSERT INTO t1 VALUES (1); |
|
|
--sync_slave_with_master |
Attachments
Issue Links
- relates to
-
-
- Closed
-
-
-
- Stalled
-
Activity
This happens because the master writes a table_map event to the binary log, but no row event.
The slave has a check that there should always be a row event if there was a table_map event, which
causes a crash.
I have fixed this for 10.3 by remembering all event types written to the log and not writing 'dummy' events to
the log. The fix is rather big so I propose the following for 10.1 and 10.2
int Query_log_event::do_apply_event(rpl_group_info *rgi,
const char *query_arg, uint32 q_len_arg)
Change
DBUG_ASSERT(!current_stmt_is_commit || !rgi->tables_to_lock);
to
#ifdef MYSQL_VERSION_ID >= 100300
DBUG_ASSERT(!current_stmt_is_commit || !rgi->tables_to_lock);
#endif
In 10.3 we have to also check that the master > 10.3.0
I have checked the code and this is ok to do.
I assume it stayed open by mistake – there had been commits in main trees (and hence in change logs) fixing it, and indeed, it's not reproducible any longer with this particular scenario in 10.3+.
I got something similar on 10.4, once:
2019-02-14
|
mysqld: /git/10.4/sql/log_event.cc:5464: int Query_log_event::do_apply_event(rpl_group_info*, const char*, uint32): Assertion `!current_stmt_is_commit || !rgi->tables_to_lock' failed.
|
190214 18:34:41 [ERROR] mysqld got signal 6 ;
|
Server version: 10.4.3-MariaDB-debug-log
|
|
|
linux/raise.c:54(__GI_raise)[0x7ff4a563ebd7]
|
stdlib/abort.c:91(__GI_abort)[0x7ff4a563ec82]
|
/git/10.4/sql/mysqld(_ZN15Query_log_event14do_apply_eventEP14rpl_group_infoPKcj+0x9f0)[0x55b7f3c60d2a]
|
sql/log_event.cc:5465(Query_log_event::do_apply_event(rpl_group_info*, char const*, unsigned int))[0x55b7f3c6029f]
|
sql/log_event.cc:5359(Query_log_event::do_apply_event(rpl_group_info*))[0x55b7f300975d]
|
sql/log_event.h:1481(Log_event::apply_event(rpl_group_info*))[0x55b7f2ff0748]
|
sql/slave.cc:3950(apply_event_and_update_pos_apply(Log_event*, THD*, rpl_group_info*, int))[0x55b7f2ff115a]
|
sql/slave.cc:4105(apply_event_and_update_pos(Log_event*, THD*, rpl_group_info*))[0x55b7f2ff2651]
|
sql/slave.cc:4425(exec_relay_log_event(THD*, Relay_log_info*, rpl_group_info*))[0x55b7f2ff8a19]
|
/lib/x86_64-linux-gnu/libpthread.so.0(+0x76ba)[0x7ff4a64876ba]
|
x86_64/clone.S:111(clone)[0x7ff4a571841d]
|
Still happens on current 10.1 and 10.2.