dovecot-1.0-1.2.rc15.el5 on CentOS 5
I reboot the server and dovecot fails to start. But I can then issue "service dovecot start" and it starts up just fine. How to debug? I don't see anything in /var/log/messages nor /var/log/maillog.
I did have to set SELinux to permissive (in /etc/selinux/config), as dovecot can't get at its SSL certs. I'm debugging that independently. (SELinux is also disabled for clamav, another item to debug.)
Around 12:02pm on Wednesday, October 24, 2007 (UK time), Kenneth Porter scrawled:
I reboot the server and dovecot fails to start. But I can then issue "service dovecot start" and it starts up just fine. How to debug? I don't see anything in /var/log/messages nor /var/log/maillog.
I know its proobably a daft question, but what does "/sbin/chkconfig --list dovecot" produce?
Steve
--On Wednesday, October 24, 2007 12:59 PM +0100 Steve Searle steve@stevesearle.com wrote:
I know its proobably a daft question, but what does "/sbin/chkconfig --list dovecot" produce?
A fair question, and the first thing I checked. ;)
[ken@segw2 ~]$ chkconfig --list dovecot dovecot 0:off 1:off 2:off 3:on 4:on 5:on 6:off [ken@segw2 ~]$ runlevel N 3
Kenneth Porter wrote:
[ken@segw2 ~]$ chkconfig --list dovecot dovecot 0:off 1:off 2:off 3:on 4:on 5:on 6:off [ken@segw2 ~]$ runlevel N 3
despite that which seems to be correct, can you please post the output of the following? find /etc/rc* -name *dovecot* -exec ls -l {} ;
ciao - Michael
On Wednesday, October 24, 2007 5:23 PM +0200 Michael Kress kress@hal.saar.de wrote:
despite that which seems to be correct, can you please post the output of the following? find /etc/rc* -name *dovecot* -exec ls -l {} ;
Looks right to me:
[ken@segw2 ~]$ [ken@segw2 ~]$ find /etc/rc* -name *dovecot* -ls 85329138 0 lrwxrwxrwx 1 root root 17 Aug 13 01:40 /etc/rc.d/rc3.d/S65dovecot -> ../init.d/dovecot 85329136 0 lrwxrwxrwx 1 root root 17 Jul 24 06:00 /etc/rc.d/rc1.d/K35dovecot -> ../init.d/dovecot 85329135 0 lrwxrwxrwx 1 root root 17 Jul 24 06:00 /etc/rc.d/rc0.d/K35dovecot -> ../init.d/dovecot 85329141 0 lrwxrwxrwx 1 root root 17 Jul 24 06:00 /etc/rc.d/rc6.d/K35dovecot -> ../init.d/dovecot 85329137 0 lrwxrwxrwx 1 root root 17 Jul 24 06:00 /etc/rc.d/rc2.d/K35dovecot -> ../init.d/dovecot 85329139 0 lrwxrwxrwx 1 root root 17 Aug 13 01:40 /etc/rc.d/rc4.d/S65dovecot -> ../init.d/dovecot 85329134 8 -rwxr-xr-x 1 root root 917 Mar 14 2007 /etc/rc.d/init.d/dovecot 85329140 0 lrwxrwxrwx 1 root root 17 Aug 13 01:40 /etc/rc.d/rc5.d/S65dovecot -> ../init.d/dovecot
on 10/24/2007 12:51 PM Kenneth Porter spake the following:
On Wednesday, October 24, 2007 5:23 PM +0200 Michael Kress kress@hal.saar.de wrote:
despite that which seems to be correct, can you please post the output of the following? find /etc/rc* -name *dovecot* -exec ls -l {} ;
Looks right to me:
[ken@segw2 ~]$ [ken@segw2 ~]$ find /etc/rc* -name *dovecot* -ls 85329138 0 lrwxrwxrwx 1 root root 17 Aug 13 01:40 /etc/rc.d/rc3.d/S65dovecot -> ../init.d/dovecot 85329136 0 lrwxrwxrwx 1 root root 17 Jul 24 06:00 /etc/rc.d/rc1.d/K35dovecot -> ../init.d/dovecot 85329135 0 lrwxrwxrwx 1 root root 17 Jul 24 06:00 /etc/rc.d/rc0.d/K35dovecot -> ../init.d/dovecot 85329141 0 lrwxrwxrwx 1 root root 17 Jul 24 06:00 /etc/rc.d/rc6.d/K35dovecot -> ../init.d/dovecot 85329137 0 lrwxrwxrwx 1 root root 17 Jul 24 06:00 /etc/rc.d/rc2.d/K35dovecot -> ../init.d/dovecot 85329139 0 lrwxrwxrwx 1 root root 17 Aug 13 01:40 /etc/rc.d/rc4.d/S65dovecot -> ../init.d/dovecot 85329134 8 -rwxr-xr-x 1 root root 917 Mar 14 2007 /etc/rc.d/init.d/dovecot 85329140 0 lrwxrwxrwx 1 root root 17 Aug 13 01:40 /etc/rc.d/rc5.d/S65dovecot -> ../init.d/dovecot
Are you using any auth services in dovecot that are starting after it does like sasl or ldap? Or maybe a database connection?
--On Wednesday, October 24, 2007 2:15 PM -0700 Scott Silva ssilva@sgvwater.com wrote:
Are you using any auth services in dovecot that are starting after it does like sasl or ldap? Or maybe a database connection?
Ah, good possibility. Dovecot starts at priority 65, while saslauthd starts at 95. Shouldn't this only be a problem once someone attempts a login? Perhaps it's a Dovecot bug.
Kenneth Porter schrieb:
Ah, good possibility. Dovecot starts at priority 65, while saslauthd starts at 95. Shouldn't this only be a problem once someone attempts a login? Perhaps it's a Dovecot bug.
My mail related stuff looks and works like this: lrwxrwxrwx 1 root root 17 Sep 18 2007 S65dovecot -> ../init.d/dovecot lrwxrwxrwx 1 root root 17 Sep 18 2007 S80postfix -> ../init.d/postfix lrwxrwxrwx 1 root root 22 Sep 18 2007 S80spamassassin -> ../init.d/spamassassin lrwxrwxrwx 1 root root 19 Sep 18 2007 S95saslauthd -> ../init.d/saslauthd lrwxrwxrwx 1 root root 25 Sep 18 2007 S99pop-before-smtp -> ../init.d/pop-before-smtp
But, I don't use mysql for storing any information.
Greetings - Michael
Kenneth Porter wrote:
dovecot-1.0-1.2.rc15.el5 on CentOS 5
I reboot the server and dovecot fails to start. But I can then issue "service dovecot start" and it starts up just fine. How to debug? I don't see anything in /var/log/messages nor /var/log/maillog.
I did have to set SELinux to permissive (in /etc/selinux/config), as dovecot can't get at its SSL certs. I'm debugging that independently. (SELinux is also disabled for clamav, another item to debug.)
This is something I am having problems with using USB2 connected external storage.
The USB filesystems are not initialized when it is time for mount ... therefore I need to do a noauto in /etc/fstab and mount them later in the boot process.
This caused httpd to not start (as there was a directory used as the DocumentRoot for a Virtual Machine on the USB drive).
Is it possible that you are mounting a filesystem after the postfix tries to start ... and it is available after startup but not at init.d/postfix start time?
Johnny Hughes wrote on Wed, 24 Oct 2007 07:17:06 -0500:
Is it possible
And if not: do you watch the console when it is booting up, do you see a delay when Dovecot is starting up (and failing)?
Kai
On Wednesday, October 24, 2007 4:49 PM +0200 Kai Schaetzl maillists@conactive.com wrote:
And if not: do you watch the console when it is booting up, do you see a delay when Dovecot is starting up (and failing)?
Good idea, I'll try that in the morning (when the machine isn't in use).
On Wednesday, October 24, 2007 7:17 AM -0500 Johnny Hughes johnny@centos.org wrote:
Is it possible that you are mounting a filesystem after the postfix tries to start ... and it is available after startup but not at init.d/postfix start time?
I do have a USB drive attached, but it's not mounted. (I mount it when I need to run a filesystem dump to back up the server, to /mnt/sdc.)
But I do suspect a similar kind of race is involved.
-
Johnny Hughes -
Kai Schaetzl -
Kenneth Porter -
Michael Kress -
Scott Silva -
Steve Searle