Bug #14106 IM: im_life_cycle and im_utils tests fail on FreeBSD
Submitted: 18 Oct 2005 1:29 Modified: 24 May 2006 19:26
Reporter: Petr Chardin Email Updates:
Status: Closed Impact on me:
None 
Category:Instance Manager Severity:S3 (Non-critical)
Version:5.0 OS:
Assigned to: Alexander Nozdrin CPU Architecture:Any

[18 Oct 2005 1:29] Petr Chardin 
Description:
mentioned tests fail on FreeBSD

TEST                           RESULT
-------------------------------------------------------

im_life_cycle                  [ fail ]

Errors are (from /usr/home/mysqldev/petr/mysql-5.0.15/mysql-test/var/log/mysqltest-time) :
mysqltest: At line 132: query 'START INSTANCE mysqld2' failed: 3002: The instance is already started
(the last lines may be the most important ones)

This happens since IM cannot kill the instance. Something strange happens with kill there: kill(SIGTERM) only forces an instance to remove the pidfile, but the server is still alive. kill(SIGKILL)  issued by IM right after that doesn't work. However if IM just always do kill(SIGKILL), instances are killed fine.
Looks like a FreeBSD thread issue.

How to repeat:
run tests
[5 May 2006 13:20] Bugs System 
A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/6012
[6 May 2006 9:58] Bugs System 
A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/6046
[6 May 2006 10:54] Konstantin Osipov 
The patch is OK to push.
[10 May 2006 10:33] Bugs System 
A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/6187
[18 May 2006 21:16] Dmitry Lenev 
Fixed in 5.0.22 and 5.1.11
[24 May 2006 19:26] Paul DuBois 
Noted in 5.0.23, 5.1.11 changelogs.

Added the <option>--angel-pid-file</option> option to
<command>mysqlmanager</command> for specifying the file in
which the angel process records its process ID when
<command>mysqlmanager</command> runs in daemon mode. 

The bug involves test cases, but the fix involves a new option
(--angel-pid-file), so I've added that to the mysqlmanager section
in the manual.