Zimbra slow in very small deployment

[xeon]

You will get better performance with 1 vCPU. I ran Zimbra on VMware Server 1.x on Ubuntu 8.04 host and guest with two 160GB SATA drives in Linux RAID1 with 5 other VMs on the machine (two Windows XP, Windows 2003 DC, and two Ubuntu). Zimbra worked great after I disabled the logger. I saw major performance issues when logger was enabled because of how many disk IOs it uses. Disk IO seems to always be the performance issue with virtual machines. My box only had 6GB of RAM, 2.33 GHz Quad-Core XEON, and the Zimbra VM had 1 vCPU with 1.5GB of RAM. The RAM was preallocated and no swapping of memory. There are a lot of tuning you need to do on the VMware/host side to have good performance on your virtual machines.

PS: I used Ubuntu 8.04 JeOS (32-bit) as the guest OS.

[alberto]

I have not check restarting just the zimbra server. Your install is almost exactly like mine.
I will try what you said and restart just the server. Lately I have notice that restarting the server takes a long time too.
I take it that all you do is sudo /etc/init.d/zimbra restart right?

[xeon]

I take it that all you do is sudo /etc/init.d/zimbra restart right?

su - zimbra
zmcontrol stop
Wait ...
Check for remaining process; ps waux | grep zimbra
zmcontrol start

[alberto]

An update on the test.
Zimbra stopped correctly and there were not remaining processes, at least this time around. The machine had been running for just a day so it was not the worst case scenario.
How can one disable the logger? I would try it but what i don't understand is that if that were the issue it should happen all the time rigth? Why does it happen only after some time running?

[xeon]

I beleive this is what you need to do, I dont have my documentation in front of me to check for sure:

su - zimbra
zmprov ms mail.domain.com -zimbraServiceEnabled logger
zmcontrol stop
Wait ...
Check for remaining process; ps waux | grep zimbra
zmcontrol start

Then do "zmcontrol status" to see if its running or not.