every 2 minutes 100% CPU usage

[stringi]

Hello.
Every 2 minutes I have 100% CPU usage in my Zimbra box.

Htop shows that 100% generates "/opt/zimbra/java/bin/java -XX:ErrorFile=/opt/zimbra/log -client -Dzimbra.home=/opt/zimbra -cp /opt/zimbra" process.

Zimbra 6.0.2 is running on virtual (VMWare Infrastructure 3.5) Centos5.4

thanks for help...

[mtk]

I have exactly the same problem running the Open Source Edition 6.0.2 32bit on a Debian 5 running in VirtualBox.

The system has been installed from scratch.

Anybody has a clue whats going wrong there?

Thanks

[uxbod]

How much memory has been allocated to the virtual machine ?

[mtk]

1,5GB - unfortunatelly I was not able to assign 2GB to the VM. 1,5 seems to be the max for my machine.

[uxbod]

How many users ? What is the profile of the server eg. ZWC/IMAP ? Would you post some stats from the server eg. free mem

[mtk]

The machine is totally set up from scratch. The problem already occures with no Zimbra users set up at all, even before I log on to the console the first time.
There are no mailboxes existing besides the ones coming with the setup.
The server has been installed with the standard options.
However, same situation with one mailbox and around 1GB of data.

I installed the machine a second time to see wether there was something wrong with the Debian or Zimbra-setup or anything else. However, same issues. So it seems that I can reproduce it. I installed the server exactly like I did it with previous 6.0.x-versions. Earlier versions didn't show the same problem.

There is not much free RAM remaining. With no mailboxes, more then 1GB is occupied. I think it's around 100-200MB of RAM free, but I would need to check this later when I'm in front of the machine.

Please let me know when you need more details.

[Mistoffeles]

Originally Posted by mtk

There is not much free RAM remaining.

There is your problem right there.

My server has 4GB of RAM and is running a clean install of RHEL 5.3 64-bit and Zimbra 5, with absolutely nothing else major running on it, and nothing under it such as VMware, VirtualBox or what have you.

This configuration consumes just a smidge under 2.5GB consistently, with only a handfull of accounts created and very few actually in use. This is with no GUI (/etc/inittab id:3:initdefault: ) running at all.

Remember, the production requirements of Zimbra 5, let alone 6, are minimum 2GB and recommended 4GB, with 64-bit OS on large deployments recommended.

Note: Check your SELinux settings, shut that bugger off completely if it is running.

[mtk]

I doubt that, because problems seem to occure during startup at a time the system has more then 70% of free RAM remaining.

My (blank) installation needs around 1GB with started ZCS, so there is still enough space remaining.

Also: It worked with 6.0.1 and exact same hardware configuration flawless.

SELinux is not installed.

[raj]

try the solution i posted here:

High CPU spikes every 1 minute

Raj

[mmorse]

Stab in the dark till you attach more zmstats/zmdialog (carefull you don't post your localconfig passwords) but Bug 42715 – zmmtaconfig spawned java process consumed high cpu