Hi all,
i've been trying to find out whats happening in my dovecot setup every weekend.. with no luck.
I have postfix/dovecot/procmail/spamassassin/clamd setup running, and all is good except dovecot seems to crash every weekend. Postfix and all other programs run with no probs.
Now, i think it might be some weekend job that crashes it, but no idea where.
Anyone know any weekly dovecot related jobs (cron maybe) that might do this?
I have 3 servers with same config crashing pretty much at the same time. (CentOS Linux 5.3, Webmin version: 1.470, Virtualmin version: 3.68 Pro)
You'd want to look in the cronjobs that happen on the weekend (Webmin has a module for cronjobs, or you can look in /etc/cron*).
You should also check the maillog during the period where the problem occurs. You can grep for dovecot to get rid of all the extra stuff (maillog is very chattery).
--
Check out the forum guidelines!
Yeah i've been trying to find out which job it might be..
/etc/cron.weekly/ has 2 jobs:
-rwxr-xr-x 1 root root 380 Mar 28 2007 0anacron
-rwxr-xr-x 1 root root 414 Jan 6 2007 makewhatis.cron
and doubt its either one that is doing this...
the maillog is pain in the ...
[root@web01 /]# grep dovecot /var/log/maillog |wc -l
157914
:-)
i'll try to find out more info..
- Yorkki
You also have the root crontab that could be running jobs on the weekend.
You could try getting more specific in your grepping.
grep dovecot /var/log/maillog | grep -i fatal
And try "error" or "shutting down" instead of fatal, perhaps. I dunno. I'm not sure what to look for, but I'm sure there's <i>something</i> in the log about it.
--
Check out the forum guidelines!
Found this:
May 17 23:59:47 web01 dovecot: Time just moved backwards by 13 seconds. This might cause a lot of problems, so I'll just
kill myself now. http://wiki.dovecot.org/TimeMovedBackwards
...hmm ... i have a ntp update setup in cron...
- Yorkki
Not sure if there is a better solution but i can add "&& service dovecot restart" after ntpdate job i guess...
- Yorkki
The better solution is to run ntpd, instead of running ntpdate once per week.
ntpd updates time constantly (and keeps your clock accurate to less than a second).
service ntpd start
chkconfig --level 345 ntpd on
--
Check out the forum guidelines!
Ok i'll give it a try,
thanks Joe,
- Yorkki
webmin has a module for it under hardware-system time
i run the sync every 20 minutes.