On 25. jul. 2007, at 17.40, Nick Upson wrote:
> Thanks, so if I always do a reload after the logrotate I can avoid
> the bug?
No, if you do a monit reload you will encounter the bug. The
confusing bit here is that calling monit reload is the correct thing
to do after a logrotate, but will provoke the bug. Update to the CVS
version or quit and restart monit after logrotate.
> On 25/07/07, Jan-Henrik Haukeland <address@hidden> wrote:
>> On 25. jul. 2007, at 16.20, Nick Upson wrote:
>>
>> > I have defined a log for monit to write and it does but when
>> logrotate
>> > moves the first log aside monit continues to write to the old file
>> > with the new name.
>> >
>> > I tried doing a "monit reload" which usually works but sometimes
>> monit
>> > seems to get confused and stop writing to any logfile at all.
>> >
>> > Can someone please tell me what I'm doing wrong
>>
>> Unfortunately there is a bug in monit 4.9 that after a monit reload
>> has the log function use an uninitialized mutex. Sounds like you
>> found that bug :) It is fixed in CVS and you can try this snapshot
>> version of monit:
>>
>> http://tildeslash.com/monit/dist/snapshot/monit-4.10-cvs.tar.gz
>>
>> Otherwise, after a logrotate, calling 'monit reload' is the correct
>> way to do this. This way, monit reopens its log file. If you don't
>> call 'monit reload' monit will continue writing to the old log file
>> even if it was renamed.
--
To unsubscribe:
http://lists.nongnu.org/mailman/listinfo/monit-general