I run a weekly off-line backup of the libvirt VM container that my Zimbra 8.0.3 FOSS installation runs on top of. A backup script on the main host where the Zimbra VM resides is initiated out of cron to run an libvirt API call to shut down the guest gracefully and leverages the acpid capabilities of the Zimbra VM guest OS (RHEL 6.4 64bit). The script also verifies that guest has been cleanly shut down. Thereafter, the script ensures the VM's disk image and libvirt configuration is backed up. After the backup completes, the same script starts the VM up again, and given that the Zimbra installation is configured to start with the OS via chkconfig, it will start up automatically too.

The problem I'm experiencing is that infrequently the Postfix appears to have been started, but will not bind to port 25. There are no other MTAs installed, other than sendmail (due to other RPM dependencies), but sendmail is configured via chkconfig to not start at any runlevel. I see in the logs that zmcontrol at least attempts to start the MTA:

Jun 5 03:11:46 redacted zimbramon[1464]: 1464:info: Starting mta via zmcontrol

When this Postfix issue shows itself, I can't find anything useful in the logs to point to why it did not start. I looked in the /var/log/zimbra.log file, as well as in /var/log/messages and there are no hints of "could not bind to 0.0.0.0:25" or other errors that would be worth investigating. I've also looked through /opt/zimbra/log, but there does not to be anything enlightening in there either relating to zmcontrol.

The only signs of this condition are the following:
1) In /var/log/messages, the following is logged:
Jun 5 03:11:46 redacted zimbramon[1464]: 1464:info: Starting mta via zmcontrol

But in cases where Postfix doesn't successfully start, the following log entries *do not appear* (neither do any specific errors as to why Postfix may not have been able to start):
Jun 5 14:22:49 redacted /postfix-script[2849]: warning: not owned by root: /opt/zimbra/postfix-2.10-20121007.2z/conf/main.cf
Jun 5 14:22:49 redacted /postfix-script[2850]: warning: not owned by root: /opt/zimbra/postfix-2.10-20121007.2z/conf/master.cf
Jun 5 14:22:49 redacted /postfix-script[2852]: warning: not owned by root: /opt/zimbra/postfix-2.10-20121007.2z/conf/master.cf.in
Jun 5 14:22:49 redacted /postfix-script[2853]: warning: not owned by root: /opt/zimbra/postfix-2.10-20121007.2z/conf/tag_as_foreign.re
Jun 5 14:22:49 redacted /postfix-script[2854]: warning: not owned by root: /opt/zimbra/postfix-2.10-20121007.2z/conf/tag_as_originating.re
Jun 5 14:22:50 redacted /postfix-script[2871]: starting the Postfix mail system
Jun 5 14:22:50 redacted postfix/master[2873]: daemon started -- version 2.10-20121007, configuration /opt/zimbra/postfix-2.10-20121007.2z/conf

2) No process is listening on TCP 25 as confirmed by netstat -antpl | grep :25
3) Postfix logs to the effect of "postfix/postqueue[nnnn]: fatal: Queue report unavailable - mail system is down" every once in a while

The web framework starts fine, Webmail works, IMAP works, it's just SMTP that is in a bad state.

Restarting the host via 'reboot' or restarting Zimbra via "service zimbra restart" always fixes the issue.

The next time this issue occurs I'll have to do more poking while the system is in the state, but curious if this issue happens to anyone else?