ESX Server issue or VC perf anomaly

[Rumple]

I have the following setup

3- x366 Servers. 16GB RAM in each

ESX01, ESX02 are Single core, quad processor

ESX03, Dual core quad processor

ESX01 - 9 VM's, typically only a single VM that goes over a GHz or so of processor usage

ESX02 - 15 VM's - Only a single VM thats ever really busy but uses about 2GHz of processor

ESX03 - 7 VM's - Only single VM that typically uses 1.4GHz of processor

So, pretty underutilized hosts with lots of capacity, but while troubleshooting a VM on ESX02 I noticed an interesting graph pattern in VC.

On ESX02 and ESX03 (but not ESX01) at exactly 25 minutes past the hour 24x7, my CPU graphs drop to zero and then go back up to average CPU usage again.

From the time the CPU drops to the time its levels out again, its exactly 10 minutes. My thought at first was it was VC database rollups happening, but why is it only showing up on 2/3 ESX Servers.

[kbk00 · Accepted solution]

I don't think it's the rollup issue...

Where is your virtual center server? Is it on ESX01?

Another troubleshooting method is to just connect directly (with the VI client) to esx02 or 03 and see if the drop happens there or if it's just in VC.

[Rumple]

Good point on VC location. The VM was on ESX01, I've since migrated it to ESX01

I've looked at the ESX02 graphs directly on the box, and I am not seeing the drop in the graphs so its got to be a function of VC being a VM and getting descheduled or something.

[kbk00]

My guess is that it's something network related.

There are many installs of VC running within a VM. Yours truly is an advocate...

http://www.vmware.com/pdf/vi3_vc_in_vm.pdf