java.lang.OutOfMemoryError: Java heap space

[John Birdsey]

The Box I am testing on is a 1.6GHz intel with 1Gb RAM, 20Gb HD, 2Gb swap and is running Fedora Core 3 and the evaluation Network Version

[KevinH]

The Box I am testing on is a 1.6GHz intel with 1Gb RAM, 20Gb HD, 2Gb swap and is running Fedora Core 3 and the evaluation Network Version

The Network install is for Red Hat only. What are you doing when it runs out of memory? How big are the messages your receiving?

[John Birdsey]

There are 10 users with most having a mailbox size of 20-50 meg after 3 weeks. One user has 100Mb in the same time. I can’t see anywhere that I can easily see any message sizes or quantities. The reason I have installed the network version on fedora 3 is that I was looking at Zimbra to replace our current server configuration. We are running sendmail, Spamassassin and MailScanner serving 150 users on RedHat ES 4 and I need the added benefit of backups. But I don’t want to pay for another copy of RedHat ES just for this test.

Thanks

John

[KevinH]

There are 10 users with most having a mailbox size of 20-50 meg after 3 weeks. One user has 100Mb in the same time. I can’t see anywhere that I can easily see any message sizes or quantities. The reason I have installed the network version on fedora 3 is that I was looking at Zimbra to replace our current server configuration. We are running sendmail, Spamassassin and MailScanner serving 150 users on RedHat ES 4 and I need the added benefit of backups. But I don’t want to pay for another copy of RedHat ES just for this test.

1 GB os RAM is a little light for a prod install, but you still should be ok for a small trial. Do you get OOM exceptions frequently? Are there any other errors in the logs that may point to other problems? Anything else running on the box that may be taking up memory?

[John Birdsey]

I understand 1Gb is on the light side but i was hoping it would be enough for a 10 user test install. If i can get Zimbra stable it will replace our current box which we have just upgraded to 2Gb. I am getting OOM errors about every hour and i have to restart the mailbox service. I did have an error previously when I mistakenly ran zmbackup as root and I think it reset some of the file permissions. I tried tracking them down one by one but ended up running the setup again to get the server back operational.

I have attached my log (I have chopped some lines out and changed the email domain and server name) and it looks like the mailbox service fell over after a delete of about 1000 messages as I have seen in a previous post. But it is also falling over with little activity other then recieving emails.

Thanks again

John

[marcmac]

Is there anything else running on this box?

What's your usage profile? (Web users, POP users, IMAP, etc, msgs/day, etc.)

You mention a 1000 message delete - are you migrating to the server? That will put an abnormal load on the box.

[John Birdsey]

the box is dedicated to this test at the moment.
I have 10 users setup.
There is no pop3 or Imap access only web access.
And i think with the current issue with the mailbox service failing the users i have asked to test are not logging in.
The 1000 messages are from mail delivery subsystem from my main server of undeliverable messages where people try spamming us with guessed email addresses and forged reply address.

[marcmac]

Hmmm. In general, we should be able to handle that traffic. How was the user delete performed?

In a side note, the closer to the firewall you stop spam, the better.

[John Birdsey]

The delete was done via a empty trash as the messages are moved there with a filter. Is there a scheduled task(s) that runs, maybe making the mailbox service fail every .5 - 1 hour
I have not looked into the mail delivery subsystem messages deep enough and currently dont know how to stop the bogus messages. other then to forward the good and bad system emails to /dev/null

Thanks as always

John

[KevinH]

su - zimbra
crontab -l

Should be a few things in there. Not sure which run that frequently.