VM CPU usage maxes out after Vmotion event

[JonRoderick]

One or two VMs exhibit 100% CPU usage (generally after a DRS Vmotion event to another host) and are unresponsive, requiring a hard reboot. VMware haven't been able to see anything in the logs that might have caused this situation but I'd like to know what's going on before I get hammered by the application team for being a shyster.

Anyone point me in the direction of where I might find some answers?

Cheers.

[jrenton]

See: http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=100363...

This should have the answers you need

Regards

John

[JonRoderick]

Thanks - I remember seeing that KB when it came out.

I'm still on ESX 3.0.2 and it's not reported as applicable to that version.....that said, it's a good start so you get some points.

Cheers.

[kjb007]

What about vmware tools? Are they updated, and services running properly?

Also, is the vm and/or the service console swapping?

-KjB

[JonRoderick]

Hard to tell...need to catch it when it happens again to get that information.

As for tools, just upgraded to 3.0.2 Update 1 and in the process of upgrading alll the Tools packages but takes time on 200 VMs. That said, haven't seen issues with similar VMs.

Cheers.

[michael_webster]

I've had this happen to me today after upgrading a host to ESX 3.5 Update 4. Migrating a number of VM's off a host on ESX3.5 Update 2, to ESX3.5 Update 4, all of the VM's experience a CPU race condition and go to 100%. After going into one of the VM's the vmware-guestd service is taking up a lot of CPU. We running vCentre 2.5 Update 3, the previous problem from the KB article on this thread said the problem was meant to be fixed in vCentre 2.5 U2. Also this was to a host that is not part of our DRS Cluster. Any ideas would be greatly appreciated.

[ufo8mydog]

Hi Michael,

We are having the same issue. This is on the latest ESX and VC (U4). Basically on vmotion the guest sometimes (not all the time) starts to chew CPU. Load goes from 1-2 to 60-80. Sometimes it recovers after 5-10 minutes, other times it needs a hard reboot.

Can anyone else offer any suggestions :smileygrin: ?:|

[RLI27]

Hi all

I see the same here in my environment ESX 3.5 / VC 2.5 both on Update 4 after vMotion.

I updated the all the hosts yet but none of the VMware Tools - could there be a relation? I'm going to update the VMware Tools but this will take a while for all VMs...

Others seeing this issue? Any workarounds?

Cheers, Roger

[ufo8mydog]

Upgrading the vmtools wont make a lick of difference, I'm afraid. The issue is probably caused by the vmmemctl process going out of control, perhaps it reacts with recent Redhat/Centos kernels badly.

VMWare support (APJ region) had not the slightest idea what the problem is, how to diagnose it, or how to fix it. I don't even think they read my support ticket properly. This was after several weeks of painful wrangling. They claimed the logs were 100% ok.

We are not using memory oversubscription on any host.

Personally I gave up, turned off DRS, and am hoping for a beter experience in Vsphere4 when I trial it a litle later on!