MySQL Bugs: #27218: Simultaneous Crashes on both servers in Master-Master Replication

Bug #27218	Simultaneous Crashes on both servers in Master-Master Replication
Submitted:	16 Mar 2007 15:25	Modified:	9 May 2009 6:45
Reporter:	Jeff Freund	Email Updates:
Status:	Duplicate	Impact on me:	None
Category:	MySQL Server	Severity:	S1 (Critical)
Version:	5.0.27-standard-log	OS:	Linux (CentOS x86_64 2.6.9-22.0.1.106)
Assigned to:	Assigned Account	CPU Architecture:	Any

Description:
Set up is master-master replication, both servers fielding active production traffic.  These servers have been running production with heavy usage for over a year and this problem has happened twice - once ~6 months ago with 5.0.21 and then again 2 days ago w/ 5.0.27.

Mysql on the first server crashed.  Ten seconds later, Mysql on the second server crashed.  Both databases then struggled to come back online, repeatedly crashing during the startup processes - after about 6 retries each they finally came back online.

Resolved Backtrace on several of the crashes (some did not produce a backtrace):
resolve_stack_dump -s /tmp/mysqld.sym -n mysqld.stack
0x72657375736d63 _end + 1958367195

Attached are the logs from both servers and my.cnf files.  Binlogs are archived and can be provided for this time period.

Thanks,
Jeff

How to repeat:
Have not been able to reproduce in testing

Thank you for the report.

Please provide output of SHOW SLAVE STATUS \G and part of 'mysql-bin.000190' located on second server at position 15832169 (or 'mysql-bin.000394' at position 948093 on first).

Thank you for the feedback.

Please also provide output of mysqlbinlog --end-position 15832169 mysql-bin.000190 | tail -n 100

Please provide output of SHOW CREATE TABLE puburl, SHOW TABLE STATUS LIKE 'puburl', SHOW CREATE TABLE puburlobj and SHOW TABLE STATUS LIKE 'puburlobj' statements

Thank you for the feedback.

I can not repeat simultaneous failures on my side.

Please check if crashes are simultaneous, i.e. caused by SQL thread and not just repeatable failures on both servers?

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".

If I understood provided information correctly crashes more likely were not simultaneous. So I'll close the bug report as "Can't repeat". If you argue and can provide information which could confirm crashes were simultaneous feel free to reopen the bug report.

Apparently a duplicate of bug #29856 which fixes a crash with pattern that matches a crash dump in the associated support issue and the pre-requisite of having the binary log turned on. Setting as duplicate instead of can't repeat.