Constant crashes, connection refused on Tomcat restart

marcmac · Post by **marcmac** » Wed Feb 15, 2006 7:12 pm

Virtual server?
How much disk? (Is it full?)
We log the status so often since that's what updates the status page in the admin console.

Interkey JeffG · Post by **Interkey JeffG** » Thu Feb 16, 2006 12:59 am

[quote user="marcmac"]Virtual server?
How much disk? (Is it full?)
We log the status so often since that's what updates the status page in the admin console.[/QUOTE]

It is a physical OS install, not a Xen VM or Virtuozzo VPS, and the /opt partition it's installed upon is 6 GB with 4.5 GB free. The filesystem is ext3, and not running through LVM/RAID.
It would be helpful if a future revision could log the status to MySQL or somewhere else so it stops polluting the syslog and making it impossible to track down real problems.

marcmac · Post by **marcmac** » Thu Feb 16, 2006 8:30 am

think multiple servers - local syslog collection point.
Syslog is then parsed, and the information dumped into sql.
That's what we're doing.
If it's in the way of viewing the logs: grep -v STATUS

marcmac · Post by **marcmac** » Thu Feb 16, 2006 8:34 am

Can you send us the /var/log/zimbra.log, /var/log/messages, and /opt/zimbra/log/zimbra.log for about 200 lines before and after the crash?
Also - the status logging should be logged as local0.info - so if it's going to /var/log/messages, editing your syslog.conf should change that.

Interkey JeffG · Post by **Interkey JeffG** » Wed Feb 22, 2006 3:11 pm

[quote user="marcmac"]Can you send us the /var/log/zimbra.log, /var/log/messages, and /opt/zimbra/log/zimbra.log for about 200 lines before and after the crash?[/QUOTE]

I'll have the full logs for you shortly (it's very time consuming to type them from a PuTTY session across a terminal services connection) but I've perused them thoroughly and the only thing in the logs at all other than status messages from within a few minutes of the most recent crash (as detected by Nagios, since there's nothing to indicate any problem otherwise) is a notice from amavis about blocked spam and a single email message coming into Postfix. I can't imagine what either of these would have to do with Tomcat crashing. Other crashes have nothing at all.
Incidentally, I'm not thrilled that Tomcat's functioning as a single point of failure for all mail receipt services in Zimbra, whether through POP3, IMAP or the web interface.

marcmac · Post by **marcmac** » Wed Feb 22, 2006 3:12 pm

is zimbraMtaAuthHost zimbraMtaAuthUrl and zimbraMtaMailMode set on this server?

Dux T · Post by **Dux T** » Sat Feb 25, 2006 1:26 pm

Linux wifi.duxarea.com 2.6.9-22.0.2.ELsmp #1 SMP Tue Jan 17 07:10:04 CST 2006 i686 i686 i386 GNU/Linux
zcs-3.0.1_GA_160.RHEL4
CentOS 4 3gHz P4 with 512MB Ram 80GB hdd, this machine is dedicated to zimbra
I'll grab and post the logs when I can. The server seems to crash when users are doing things in their folders from outlook ... one user in particular has a folder with 1500+ emails in it and it crashes while downloading headers... his other folder with 3500+ messages doesn't seem to have a problem.
I was thinking it was something outlook was doing but whe we tried Thunderbird the same thing happens ... the web client works fine. Of course it could all be coincedental.
Here are the entries from /opt/zimbra/log/zimbra.log around the time of several crashes

phoenix · Post by **phoenix** » Sat Feb 25, 2006 2:42 pm

[quote user="Dux T"]Here are the entries from /opt/zimbra/log/zimbra.log around the time of several crashes[/QUOTE]

I think you forgot something there.

Dux T · Post by **Dux T** » Sat Feb 25, 2006 4:34 pm

whoops - missed the file too large message on the attachment popup Here it is
zimbra.log.partial.zip

bobby · Post by **bobby** » Sat Feb 25, 2006 5:02 pm

did you kill -3 when it hangs and check catalina.out?