Dovecot freezes

[dennisc]

Sometimes several times a day I get logs like these:

May 31 14:07:01 hostname dovecot[89171]: pop3-login: Disconnected: Inactivity: rip=127.0.0.1, lip=127.0.0.1, secured
May 31 14:08:01 hostname dovecot[89171]: pop3-login: Disconnected: Inactivity: rip=127.0.0.1, lip=127.0.0.1, secured

on one of our servers. Nothing can connect to Dovecot anymore when that happens. After a Dovecot restart it operates normal again.

Anyone?

[Dillard]

*BUMP*

I've exactly the same issue (on FreeBSD 6.0 BTW).

It works for a few days, then 'dies' unexpectly. The only difference in the log-file (dovecot.log) is that now it's states:

Sep 03 19:02:31 Info: imap-login: Disconnected: rip=xxxxxxxxx, lip=xxxxxxxxxxx

while the disconnected-remark isn't there when it's working normally.

All the daemons seems to operate normally, when I try a manual connection on port 143 i receive to following:


strauss# telnet localhost 143
Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
* OK Dovecot DA ready.
user okg123
* OK Waiting for authentication process to respond..
pass xxxxxx
* OK Waiting for authentication process to respond..
list
* OK Waiting for authentication process to respond..
exit
* OK Waiting for authentication process to respond..
quit
* OK Waiting for authentication process to respond..


the only conclusion I can draw from this is that the auth-daemon (dovecot-auth) is dying, restarting dovecot solves the issue temporarily.

Dovecot version = 1.0.rc7

Anybody any ideas to solve this?

[andyl]

I've seen the same thing; my way of "solving" it has been to have Nagios automatically restart Dovecot when it can't connect.

I'm on 1.0rc7 now (and FreeBSD 6.0), and I saw it with earlier versions as well.

I've only seen the problem once every few days, and it seems to be load-related. I did adjust the setting

auth_worker_max_count = 60

from

auth_worker_max_count = 30

and haven't seen the problem since I made the change, but that might just be because I haven't waited long enough for the problem to happen again.

Andy

[Dillard]

andyl,

Thanks, I will give this setting a try, you never know.

I added:

auth_process_size = 256

perhaps some good will come out of it.

[Mitsurugi]

Sometimes several times a day I get logs like these:

May 31 14:07:01 hostname dovecot[89171]: pop3-login: Disconnected: Inactivity: rip=127.0.0.1, lip=127.0.0.1, secured
May 31 14:08:01 hostname dovecot[89171]: pop3-login: Disconnected: Inactivity: rip=127.0.0.1, lip=127.0.0.1, secured

on one of our servers. Nothing can connect to Dovecot anymore when that happens. After a Dovecot restart it operates normal again.

Anyone?

I'm getting this exact same problem, it's happened several times today. Please help urgently, our users are complaining that their email keeps going down.

If it's load-based, how can I fix it?