Bug #38289 rpl_temporary fails sporadically on windows: server has gone away
Submitted: 22 Jul 2008 17:28 Modified: 9 Feb 2009 10:12
Reporter: Sven Sandberg Email Updates:
Status: Can't repeat Impact on me:
None 
Category:Tests: Replication Severity:S7 (Test Cases)
Version:6.0-bugteam OS:Any
Assigned to: Serge Kozlov CPU Architecture:Any
Tags: crash, pushbuild, rpl_temporary, server has gone away, sporadic, test failure

[22 Jul 2008 17:28] Sven Sandberg
Description:
rpl_temporary fails on windows hosts in pushbuild with this message:

rpl.rpl_temporary 'mix'        [ fail ]

mysqltest: At line 156: failed in 'select master_pos_wait('master-bin.000001', 2863)': 2013: Lost connection to MySQL server during query

The result from queries just before the failure was:
< snip >
SET TIMESTAMP=1040324211;
SET @@session.pseudo_thread_id=1;
drop temporary table t3;
SET TIMESTAMP=1040324219;
SET @@session.pseudo_thread_id=2;
insert into t2 select count(*) from t3;
SET TIMESTAMP=1040324224;
SET @@session.pseudo_thread_id=2;
drop temporary table t3;
select * from t2 ORDER BY f;
f
5
7
drop table t1,t2;
create temporary table t3 (f int);
create temporary table t4 (f int);
create table t5 (f int);
stop slave;
insert into t5 select * from t4;
start slave;

More results from queries before failure can be found in e:\pb\bzr_mysql-6.0-bugteam\67\mysql-6.0.7-alpha-pb67\mysql-test\var-n_mix-5\log\rpl_temporary.log

Stopping All Servers
Restoring snapshot of databases

How to repeat:
https://intranet.mysql.com/secure/pushbuild/showpush.pl?dir=bzr_mysql-6.0-bugteam&order=67 vm-win2003-64-b/n_mix, vm-win2003-64-b/ps_stm, win2003-x86/n_mix

Suggested fix:
May be related to BUG#38269.

This bug is impossible to analyze since no server logs are saved in pushbuild.
Awaiting either fix to BUG#38269 or fix to BUG#38181.
[8 Oct 2008 15:44] Lars Thalmann
The bug that this is waiting for has been fixed.
Set to "open".
[19 Dec 2008 9:55] Sven Sandberg
Still happens in pushbuild, see xref: http://tinyurl.com/4gcels

This happens very seldom, but when it happens, it seems to happen for all runs of the test in the same push. Looks like host problems.