Zimbra stoped working

[marcmac]

Change the ldap values for zimbraMailPort and zimbraMailSSLPort:
zmprov ms <servername> zimbraMailPort <port> zimbraMailSSLPort <sslPort>

then restart tomcat and your server will run on the correct ports.

[Isi]

You guys/gals haven't led me astray yet. The above commands did fix my problem. When was this added? I have been doing the manual modifications as described in other posts since M2 (didn't use M1) and that seemed to work fine.

Are there other options that can be passed into this tool such as an IP Address to bind to

Thanks again for the expedient help,

-- Isi

[marcmac]

We're always adding new stuff - the ability to change the web ports is new for M4 (from multiple requests in the forums).

We don't (yet) have the ability to specify a bind address, yet.

[ccmacmil]

Originally Posted by marcmac

Not known yet - but we've got the log file, and we're looking at it.

I don't know if this is related, but I have a similar problem with tomcat starting and failing with the only resolution I can find being "chown zimbra.zimbra /opt/zimbra/redolog/redo.log" - otherwise, I get "mailbox stopped" status upon restart.

This did not happen until I upgrades from Beta to GA OSS. Any comments?

[anand]

http://www.zimbra.com/forums/administrators/3418-mystery-solved-unclean-shutdown-causes-redo-log-owned-root-root.html#post19505

[ahhhh]

I'm by no means a linux or zimbra administrator, but I'm trying to help out since our linux/zimbra guy is on a leave of absence at the moment. We are running Release 3.1.0_GA_332.DEBIAN3.1 DEBIAN3.1 FOSS edition on a P4 2.53 with 2GB RAM, with about 6 active IMAP users with Thunderbird and 2 with Outlook 2000, one of which uses IMAP and POP (with leave messages on server) simultaneously.

Since yesterday, we've been having a lot of problems. After much rebooting, and stopping and starting of zimbra using zmcontrol, I have found that it is only tomcat that is dying, and seems to be falling down a few times an hour. I can get it back up by using /opt/zimbra/bin/tomcat start, but I know that's a dirty way to do it, and it sure doesn't last very long.

The only error I can find so far is the following error in the file /opt/zimbra/log/zimbra.log.

2006-08-15 14:57:46,894 FATAL [LmtpServer-5] [] system - Fatal error occurred while handling connection
java.lang.OutOfMemoryError: Requested array size exceeds VM limit

Once this happens, it doesn't listen anymore and I need to start tomcat again. Before starting, I tried ps auxww | grep tomcat, and only got

root 13958 0.0 0.0 2880 804 pts/1 S+ 16:10 0:00 grep tomcat

in the results, so tomcat seems to be stopped)

I have checked the permissions on the redo.log, and it is zimbra:zimbra. I have tried changing the zimbraMailPort and zimbraMailSSLPort using zmprov ms, and found that it lasts about the same amount of time, so I changed it back.

Is there anything else I can check?

Thanks for your time

Ahmad van der Breggen
MarkeTel Systems

[KevinH]

Originally Posted by ahhhh

Is there anything else I can check?

2GB of memory should be fine for only 6 users. The FATAL error is happening at LMTP (which is where Zimbra receives mail). Are you getting some large attachments or extremely big messages?

[ahhhh]

Originally Posted by KevinH

2GB of memory should be fine for only 6 users. The FATAL error is happening at LMTP (which is where Zimbra receives mail). Are you getting some large attachments or extremely big messages?

I checked the queue using postqueue -p, and the biggest message in the queue is 80k (there were 70 messages in the queue when I left last night, about 300 this morning since it was dead again overnight)

[ahhhh]

Should I maybe try to clear out some of the spam from the deferred queue? Since it only shows the latest attempt time, I can't tell which ones have been in there for a long time. It has processed them down to 130 in the queue now, but tomcat is still failing frequently.

Sorry for being so lost here.

Ahmad

[ahhhh]

some more debugging:

I tried sending a test email to my zimbra account and my gmail account from my zimbra account, and the one to the zimbra account was deferred, but the one to the gmail account was delivered. Then I sent an email to my zimbra account from my gmail account, and it got through just fine (the one from my zimbra account is still deferred)

I also put all of the ones that looked like spam in the deferred queue on hold, but that didn't seem to help (tomcat still going down every 10-20 minutes).

Update: I tried sending another one from gmail, and this time it failed, so it seems fairly random whether mail gets through or not. I turned of antispam, antivirus, and spell check just to have less variables, but it still fails.

My last test for today: I put all deferred messages on hold to see if it is something in the queue.

I started requeuing messages from the hold area, and narrowed it down to what I believe is Outlook 2000 Calendar Sharing communications (Net Folders). The problems also started happening after one of the zimbra users became the host of one of the shared calendars (all our other calendars are shared from a user outside of zimbra). I have narrowed it down to 60 messages that I have on hold, and I have stopped the calendar sharing on the zimbra user's computer, so if it stays running overnight, you Zimbra folks might be interested in those messages.