Zimbra hangs up without warning/error messages

[knarfomat]

Hi folks,

the problem we have is kind of hard to describe since there is little information from the Logs about what makes Zimbra crash.

First the setup:

- opensource ZCS 6.0.5 on Ubuntu 8.0.4 64 bit (which is a VM provided by Open XenServer 5.5)
- imapproxy disabled, sysloging installed
- 4 GB RAM (about 80% usage)
- 2 VCPU (peakusage about 70-80%) (think that is quite high compared to the ressources we needed to run ZCS 5)
- app. 20 users

When Zimbra hangsup, the whole Virtualization crashes (-> open XenCenter and all other VMs hang up, too). I did try to locate the error in Open XenCenter but didnt find anything... BUT first Zimbra crashes and THEN the rest, so I posted the problem here ;-))

syslog log:

#
Mar 15 01:43:00 ipcmailxen zimbramon[11631]: 11631:info: zmstat
proc.csv: timestamp, system, user, sys, idle, iowait, mailbox,
mailbox-total-cpu, mailbox-utime, mailbox-stime, mailbox-totalMB,
mailbox-rssMB, mailbox-sharedMB, mailbox-process-count, mysql,
mysql-total-cpu, mysql-utime, mysql-stime, mysql-totalMB, mysql-rssMB,
mysql-sharedMB, mysql-process-count, convertd, convertd-total-cpu,
convertd-utime, convertd-stime, convertd-totalMB, convertd-rssMB,
convertd-sharedMB, convertd-process-count, ldap, ldap-total-cpu,
ldap-utime, ldap-stime, ldap-totalMB, ldap-rssMB, ldap-sharedMB,
ldap-process-count, postfix, postfix-total-cpu, postfix-utime,
postfix-stime, postfix-totalMB, postfix-rssMB, postfix-sharedMB,
postfix-process-count, amavis, amavis-total-cpu, amavis-utime,
amavis-stime, amavis-totalMB, amavis-rssMB, amavis-sharedMB,
amavis-process-:::B53DF0D6-2FCB-11DF-A7C0-434BD9A28645:::
Mar 15 01:43:00 ipcmailxen
zimbramon[11631]: :::B53DF0D6-2FCB-11DF-A7C0-434BD9A28645:::count, clam,
clam-total-cpu, clam-utime, clam-stime, clam-totalMB, clam-rssMB,
clam-sharedMB, clam-process-count:: 03/15/2010 01:43:00, system, 31.3,
3.2, 65.2, 0.3, mailbox, 0.2, 0.1, 0.1, 2308.6, 979.5, 13.0, 1, mysql,
0.0, 0.0, 0.0, 476.6, 214.1, 4.8, 1, convertd, 0.0, 0.0, 0.0, 0.0, 0.0,
0.0, 0, ldap, 0.0, 0.0, 0.0, 373.5, 59.3, 4.8, 1, postfix, 0.0, 0.0,
0.0, 46.4, 2.2, 1.7, 4, amavis, 0.0, 0.0, 0.0, 205.1, 90.6, 3.7, 11,
clam, 0.0, 0.0, 0.0, 166.4, 129.5, 1.3, 2
Mar 15 01:43:22 ipcmailxen zimbramon[11657]: 11657:info: zmstat
mtaqueue.csv: timestamp, KBytes, requests:: 03/15/2010 01:43:22, 0, 0
Mar 15 10:09:07 ipcmailxen syslogd 1.5.0#1ubuntu1: restart.
#

the last zmstat mtaqueue.csv command usually runs smoothly several times an hour...

zimbra log looks the same... no obvious errors (at least not to a newbee like me) Only thing that is strange: the log files building up quite fast (app. 10 MB a day, dont know if that is normal...)

Zimbra (and the whole open xenserver) was fine for a week! Smooth performance, etc. Only to hang up at 2 am in the morning ?!

We had a simillar crash when having only 2GB allocated to Zimbra (which wasnt enough) but now with 4GB we should be on the safe side??

Sorry for the few information I can provide, please let me know if you need more log entries...

Anyone an idea / simliar problem?

Thanks a lot!

Frank