VMware Communities > VMTN > VMware Infrastructure™ > VI: ESX 3.0 > Discussions
Up to Discussions in VI: ESX 3.0

This Question is Possibly Answered

1 "correct" answer available (10 pts) 2 "helpful" answers available (6 pts)
3 Replies Last post: Oct 9, 2008 2:51 PM by jarydj
Reply

CPU0 stuck at 100% utilization after ESX 3.5 U1 to U2 upgrade

Sep 16, 2008 6:40 AM

Click to view jk48796's profile Novice jk48796 7 posts since
Jul 13, 2008
I have 4x ESX 3.5 build 110268 hosts (Dell PE1950's) in a HA / DRS cluster. They were upgraded from U1 to U2 over the weekend. 3 of them have CPU0 getting stuck at 100% utilization. These are dual quad-core systems.

I have tried the following to try to resolve this:
  • Migrate all VMs off one of the hosts (even with 0 VMs running CPU0 has 100% utilization).
  • Uninstalled Dell OpenManage Server Administrator.
  • Restarted mgmt-vmware and vmware-vpxa services
  • Reconfigure for HA on all 4 hosts.

esxtop shows console using 99%, top shows cimprovagt at +97%.

Is this 100% CPU usage normal? It wasn't doing it before the U2 upgrade and I'm not sure what cimprovagt is.

Reply Re: CPU0 stuck at 100% utilization after ESX 3.5 U1 to U2 upgrade Sep 16, 2008 7:19 AM
Click to view Rubeck's profile Master Rubeck 542 posts since
Mar 7, 2008

Try to stop the pegasus service which runs the CIM... (service pegasus stop). You wont be able to see HW status in the VIClient when turning this off, but try to see if this is what kills it..

/Rubeck


Reply Re: CPU0 stuck at 100% utilization after ESX 3.5 U1 to U2 upgrade Sep 16, 2008 8:49 AM
in response to: Rubeck
Click to view jk48796's profile Novice jk48796 7 posts since
Jul 13, 2008

CPU0 utilization is still at 100% after stopping the pegasus service. no change in esxtop / top with console / cimprovagt.
Reply Re: CPU0 stuck at 100% utilization after ESX 3.5 U1 to U2 upgrade Oct 9, 2008 2:51 PM
in response to: jk48796
Click to view jarydj's profile Lurker jarydj 3 posts since
Aug 7, 2007
I was running into this same issue. I did the following:

1. service pegasus stop
2. kill -9 <pid> for cimprovagt (since it didn't die after I stopped the service)
3. service pegasus start

It took a minute for everything to catch back and I saw a 'python' process run up for a few minutes. But once it "finished" doing whatever it was doing, everything seems back to normal now.
Actions
Up to Discussions in VI: ESX 3.0