Bug #50480 perfschema.start_server_off crashes the server on Sol10 in 6.0
Submitted: 20 Jan 2010 15:34 Modified: 16 Jul 2010 0:51
Reporter: Alexander Nozdrin Email Updates:
Status: Unsupported Impact on me:
None 
Category:MySQL Server: Performance Schema Severity:S1 (Critical)
Version:6.0 OS:Solaris (Solaris 10)
Assigned to: Marc ALFF CPU Architecture:Any
Tags: experimental, PB, test failure

[20 Jan 2010 15:34] Alexander Nozdrin
Description:
perfschema.start_server_off crashes the server on Sol10 in 6.0:
perfschema.start_server_off [ fail ]
        Test ended at 2010-01-20 04:44:35

CURRENT_TEST: perfschema.start_server_off
mysqltest: In included file "/export/home/pb2/test/sb_0-1201218-1263956105.78/mysql-6.0.14-alpha-solaris10-sparc-test/mysql-test/suite/perfschema/include/start_server_common.inc": At line 29: query 'select * from performance_schema.EVENTS_WAITS_CURRENT' failed: 2013: Lost connection to MySQL server during query

The result from queries just before the failure was:
< snip >
mtr
mysql
performance_schema
test
select count(*) from performance_schema.PERFORMANCE_TIMERS;
count(*)
5
select count(*) from performance_schema.SETUP_CONSUMERS;
count(*)
8
select count(*) > 0 from performance_schema.SETUP_INSTRUMENTS;
count(*) > 0
0
select count(*) from performance_schema.SETUP_TIMERS;
count(*)
1
select * from performance_schema.COND_INSTANCES;

More results from queries before failure can be found in /export/home/pb2/test/sb_0-1201218-1263956105.78/mysql-6.0.14-alpha-solaris10-sparc-test/mysql-test/var-n_mix/log/start_server_off.log

How to repeat:
See PB.
[20 Jan 2010 15:34] Alexander Nozdrin
The test case has been marked experimental on Solaris in 6.0.
[27 Jan 2010 18:48] Alexander Nozdrin
It's in 6.0-codebase-bugfixing branch
(http://pb2.norway.sun.com/web.py?template=show_pushes&branch=mysql-6.0-codebase-bugfixing)

Builds: 'jorgen.loland@su... 2010-01-26 07:30:23',
'alik@sun.com-201... 2010-01-22 09:42:22'

Just search for the test case name.
[11 Feb 2010 9:51] Sveta Smirnova
Thank you for the feedback.

Verified as described using link provided.
[16 Jul 2010 0:51] Marc ALFF
Analysis

This crash happened only in 6.0-codebase (6.0)
but never in mysql-next-mr (5.5, 5.6),
and is believed to be related to the implementation of my_atomics that
was different in 6.0-codebase for solaris at some point,
due to merge issues.

Given that the 6.0-codebase is abandoned, and that the same bug does not
appear anywhere else, this can be closed.