A lot messages deferred

[luronix]

Hi!

First, i´m sorry, my English is not god so, i hope someone can understand this an help us.

We installed Zimbra a few weeks a go, and we did not problems with the service, but this week Zimbra started to send a lot of messages of our domain to deferred queue, We checked the log and found many warning messages but one of them is about the server does not has a lot space, but all the mail service (Zimbra and mailbox) just use 5% of the space designated.

When we have 900 mails in Deferred queue, we stop and start the service and all the messagesof our domain start to be delver, but 20 min. after start the same problem.

We are using Zimbra 3.1 in a Fedora Core 4

The messages that Zimbra show us are:

Sep 21 04:04:15 correo postfix/qmgr[24373]: 405472230156: to=<me@ourdomain.com>, relay=none, delay=6, status=deferred (delivery temporarily suspend
ed: lost connection with 127.0.0.1[127.0.0.1] while sending MAIL FROM)


Sep 21 07:34:01 correo postfix/smtp[8214]: 7A50322303E7: to=<me@ourdomain.com>, relay=127.0.0.1[127.0.0.1], delay=1974, status=deferred (host 127.0.0.1
[127.0.0.1] said: 421 4.3.2 Service shutting down, Can't write to mail file: No queda espacio en el dispositivo at (eval 40) line 585, <GEN3> chunk 91. (in r
eply to end of DATA command))


Sep 21 07:34:01 correo postfix/smtp[8125]: 2A72C22303EC: to=<ccalzada@inter01.lfc.gob.mx>, relay=127.0.0.1[127.0.0.1], delay=1904, status=deferred (host 127.
0.0.1[127.0.0.1] said: 421 4.3.2 Service shutting down, Can't write to mail file: Descriptor de fichero err\303\263neo at (eval 40) line 585, <GEN3> chunk 62
. (in reply to end of DATA command))

I hope someone can hellp us

Thanks for your time.

[phoenix]

Originally Posted by luronix

[127.0.0.1] said: 421 4.3.2 Service shutting down, Can't write to mail file: No queda espacio en el dispositivo at (eval 40) line 585, <GEN3> chunk 91. (in r
eply to end of DATA command))

Doesn't this mean you have no space left on the device?

[luronix]

Originally Posted by phoenix

Doesn't this mean you have no space left on the device?

Yes it is.

[phoenix]

You need to free some space on that drive then go to the admin ui and clear the deferred queue.

[luronix]

I can´t do that 'cos 85% of the messages there are valid messages.

When i restarted the Zimbra this messages are delivering, but the problem apear 20 minutes later...

I found a partition in the server whit 100% of use it is /dev/shm but i don´t know if it could be the problem.

[shanson]

That full partition with /dev/shm is certainly the problem. The same thing just happened to us.

That's the temporary directory for amavisd-new, where all attachments get unpacked for virus checking. If it's full, amavisd-new no longer accepts messages and everything gets deferred. It's a known bug, number 8081 in bugzilla.

As a workaround I increased the tmpfs partition from 150MB to 300MB. (Shutdown Zimbra, make sure the partition is unmounted, edit /etc/fstab, restart Zimbra, the partition should be automatically remounted).