cyrus-imapd spawning multiple processes

[Jeff Laine]

On Tue, Jun 02, 2009 at 10:19:01PM +0200, Erik Norgaard wrote:
> Jeff Laine wrote:
> >The troublemaker is the cyrus. It starts and opens it's sockets, but 
> >whenever
> >I try to establish a connection to pop3 or imap ports (i.e. via telnet or 
> >cyradm) cyrus goes nuts and spawns child processes in endless manner, 
> >trashing logs with this:
> >
> >>tail /var/log/debug.log
> >
> >Jun  2 23:27:35 free master[49269]: about to exec 
> >/usr/local/cyrus/bin/pop3d
> >Jun  2 23:27:35 free master[43639]: process 49269 exited, status 1
> >Jun  2 23:27:35 free master[49270]: about to exec 
> >/usr/local/cyrus/bin/pop3d
> >Jun  2 23:27:35 free master[43639]: process 49270 exited, status 1
> >Jun  2 23:27:35 free master[49271]: about to exec 
> >/usr/local/cyrus/bin/pop3d
> >Jun  2 23:27:35 free master[43639]: process 49271 exited, status 1
> >Jun  2 23:27:35 free master[49272]: about to exec 
> >/usr/local/cyrus/bin/pop3d
> >Jun  2 23:27:35 free master[43639]: process 49272 exited, status 1
> >Jun  2 23:27:35 free master[49281]: about to exec 
> >/usr/local/cyrus/bin/pop3d
> >Jun  2 23:27:35 free master[43639]: process 49281 exited, status 1
> >...
> >
> >...until I stop the master process.
> 
> It seems that when you try to connect with pop3 a child process is 
> spawned as it should, but it dies unexpectedly so a new is spawned. 
> Could it be something as simple as permissions to access the mailbox files?
> 
> Rather than relying on log files, try to launch the master manually with 
> the -D flag, it may give you a more verbose output.
> 
> >I tried both cyrus-imapd23 and 22 with default configurations but it
> >gives the same problem every time.
> >
> >I have similar configuration on 6.4-RELEASE without such troubles.
> 
> Similar but not the same? Sounds like you should review again your 
> config files, also, the versions of cyrus-imapd on your working 
> installation, is that the same as the new one? Consider posting your 
> configs in next post.
> 
> 
> BR, Erik

Thanks for your reply, Eric.
Running master in debugging mode helped to find out that I had 640 permissions 
on libdb* files. Doh.


Thanks!


-- 
Best regards,
Jeff