MySQL Bugs: #21783: Instance Manager does not exit on fatal error

Bug #21783	Instance Manager does not exit on fatal error
Submitted:	22 Aug 2006 13:35	Modified:	6 Apr 2007 14:12
Reporter:	Alexander Nozdrin	Email Updates:
Status:	Closed	Impact on me:	None
Category:	Instance Manager	Severity:	S3 (Non-critical)
Version:	5.0.25	OS:	Linux (Linux)
Assigned to:	Alexander Nozdrin	CPU Architecture:	Any

Description:
It is expected that if a program can not initialize correctly,
it exits with error code. If Instance Manager can not initialize
correctly, sometimes it continues to run, but fails to operate properly.

For example, if IM fails to bind server socket, it does not exit,
but fails to work correctly. This is quite important for test suite,
because can lead to hidden failures.

How to repeat:
1. Bind TCP socket on port 3307 (for example, start IM);
2. Start IM. Here is log:
(alik)[instance-manager]> ./mysqlmanager --defaults-file=~/.my.cnf
060822 17:16:42 loaded user alik
060822 17:16:42 loaded user alik2
060822 17:16:42 loaded user abc
060822 17:16:42 IM pid file: '/home/alik/MySQL/var/5.0-rt-im/manager.pid'; PID: 31239.
060822 17:16:42 Listener_thread::run(): bind(ip socket) failed, 'Address already in use'

Suggested fix:
Instance Manager should exit with error status if it can not initialize.

Verified as described on Linux using last 5.0.25 BK sources

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/15096

ChangeSet@1.2344, 2006-11-09 20:38:03+03:00, petr@outpost.site +3 -0
  Bug #21783  	Instance Manager does not exit on fatal error
  Return error code on fatal error (all the rest was already in place)

The bug was fixed in the patch for BUG#15934
(im_daemon_life_cycle fails sporadically).