CPU Utilization issues

[djjoshuad]

Hello,

Let me start by saying that I have been searching the web and these forums for resolution to this problem for the last few weeks. I have found many possible answers, but none of them seem to apply to my situation. Here's a description of my problem:

I downloaded the latest Zimbra and installed on a server that is also running vmware-server. The hardware is AMD X2 5600 w/ 3G RAM, running CentOS 5.2 64-bit.

Java:

[root@odessa ~]# ps aux | grep java
zimbra 4841 40.0 0.4 613016 12068 ? Rl 05:08 136:22 /opt/zimbra/java/bin/java -client -Xmx256m -Dzimbra.home=/opt/zimbra -Djava.library.path=/opt/zimbra/lib -Djava.ext.dirs=/opt/zimbra/java/jre/lib/ext:/opt/zimbra/lib/jars:/opt/zimbra/lib/ext-common:/opt/zimbra/lib/ext/clamscanner com.zimbra.cs.account.ProvUtil -l gs odessa.xxxxxxxxxxxx.com
root 12369 0.0 0.0 3764 304 ? Ss 07:41 0:00 /opt/zimbra/libexec/zmmailboxdmgr start -Xms1156m -Xmx1156m -client -XX:NewRatio=2 -Djava.awt.headless=true -XX:MaxPermSize=128m -XX:SoftRefLRUPolicyMSPerMB=1 -Xss256k
zimbra 12370 0.3 1.9 1792116 56328 ? Sl 07:41 0:38 /opt/zimbra/java/bin/java -Xms1156m -Xmx1156m -client -XX:NewRatio=2 -Djava.awt.headless=true -XX:MaxPermSize=128m -XX:SoftRefLRUPolicyMSPerMB=1 -Xss256k -Djava.io.tmpdir=/opt/zimbra/mailboxd/work -Djava.library.path=/opt/zimbra/lib -Djava.endorsed.dirs=/opt/zimbra/mailboxd/common/endorsed -Dzimbra.config=/opt/zimbra/conf/localconfig.xml -Djetty.home=/opt/zimbra/mailboxd -DSTART=/opt/zimbra/mailboxd/etc/start.config -jar /opt/zimbra/mailboxd/start.jar /opt/zimbra/mailboxd/etc/jetty.properties /opt/zimbra/mailboxd/etc/jetty-setuid.xml /opt/zimbra/mailboxd/etc/jetty.xml
zimbra 15355 76.6 0.3 611012 11796 ? Rl Nov10 1083:46 /opt/zimbra/java/bin/java -client -Xmx256m -Dzimbra.home=/opt/zimbra -Djava.library.path=/opt/zimbra/lib -Djava.ext.dirs=/opt/zimbra/java/jre/lib/ext:/opt/zimbra/lib/jars:/opt/zimbra/lib/ext-common:/opt/zimbra/lib/ext/clamscanner com.zimbra.cs.account.ProvUtil -l gamau
zimbra 28050 40.7 0.3 581760 9864 ? Rl 04:50 146:05 /opt/zimbra/java/bin/java -client -Xmx256m -Dzimbra.home=/opt/zimbra -Djava.library.path=/opt/zimbra/lib -Djava.ext.dirs=/opt/zimbra/java/jre/lib/ext:/opt/zimbra/lib/jars:/opt/zimbra/lib/ext-common:/opt/zimbra/lib/ext/clamscanner com.zimbra.cs.account.ProvUtil -l gacf
zimbra 28339 84.8 0.3 643400 8984 pts/0 Rl Nov09 2239:01 /opt/zimbra/java/bin/java -client -Xmx256m -Dzimbra.home=/opt/zimbra -Djava.library.path=/opt/zimbra/lib -Djava.ext.dirs=/opt/zimbra/java/jre/lib/ext:/opt/zimbra/lib/jars:/opt/zimbra/lib/ext-common:/opt/zimbra/lib/ext/clamscanner com.zimbra.cs.account.ProvUtil -l gacf

This server should be easily capable of running this software, especially since there is very little email passing through the system (less than 100 messages per day, currently). Is there someplace I should start digging?

Thanks!

[uxbod]

Welcome to the forums

So what is the actual issue ? could you also post which version of ZCS you are using

su - zimbra
zmcontrol -v

[djjoshuad]

Thanks for the welcome

Please forgive my ignorance of this product. I know plenty about email architecture and several other MTAs, Linux, Tomcat, etc. etc... but this is my first time using Zimbra.

Here's the output of zmcontrol -v:

Release 5.0.10_GA_2638.RHEL5_64_20081003122014 CentOS5_64 FOSS edition

The actual problem seems to be related to clamscanner processes kicked off by zmprov (kicked off from cron, maybe?). These java processes appear to be either looping or simply taking a very long time to run. They stay alive for days, clogging up the system. You can see the one from Nov. 9th in my original post.

The workaround right now is to periodically stop all zimbra processes nicely, then kill -9 the stubborn ones. This will return the system to functional for a day or so.

[djjoshuad]

Does anyone have any ideas?

I'm wondering - clamscanner is antivirus, yes? If I turn off antivirus scanning, will I prevent clamscanner from running?

[djjoshuad]

I continue to have CPU Utilization issues. In fact, about 4-5 times daily, the IMAP server will start refusing connections. This is my cue to log in and kill a clamscanner process.

any ideas? This isn't a viable solution for me if I have to log in and kill a process 4-5 times per day. I really want to make zimbra work... should I just turn off virus scanning?

[jrefl5]

djjoshuad

I'm running 2 intel processors and 4G memory handling <50 accounts, and about 2300 emails per day with no problems.

I'm discarding about 300 e-mails/day, rejecting 12,000, and sending about 2300 through spamassassin and clamav this results in about 350 "clean" e-mails.

[djjoshuad]

while that certainly sounds like a successful implementation, I am not experiencing similar success.

Does anyone have any ideas that might help?

[xeon]

Have you looked IO on your drives? If you are running in a VM, IO is very important. I turned off my logger service and it hits the hard drive much less, which keeps the IO open for my other VMs.

Try this command to watch IO usage on your drives; "iostat -mdx 3".

The last column will be IO % on the drive. If its anywhere near 100% for a long period of time, your CPU will be very high as the VMs are waiting to hit the drive.

Hope this helps.

[phoenix]

Originally Posted by djjoshuad

while that certainly sounds like a successful implementation, I am not experiencing similar success.

Does anyone have any ideas that might help?

Is your installation of Zimbra inside the VM? How much RAM is allocated to the Zimbra server (not how much is on the physical server)? How many users are on this server and are they using the Web UI or IMAP or POP3?

[tgx]

You might also take a look at the 'top' command during these high
loads. You'll be able to see which process is parking on the CPU.