host processor status alarm

[nave]

Hello,

We've started recieveing "host processor status changed from green to red" alarms occasionally in vSphere. When I examine what triggers the alarm, it just says that it triggers an Alert when the processor status is "red". What makes the processor status red? Is that a hardware setting, or something I can configure in VMware? I do see CPU spikes when that alarm is triggered, but the CPU usage is only going from 10% to about 25%. The host is a Dell M610.

Thanks.

[AndreTheGiant]

Check alarm definition, in vCenter server.

You can see that the trigger is for 75 (yellow) and 90 (red) for 5 min lenght.

Andre

[nave]

Actually, I do not see that in the alarm definition. I've attached two screenshots.

Am I just looking in the wrong place?

[AndreTheGiant]

Sorry. I thought that was a CPU usage alarm.

In your case is very a strange issue. You have to use vendor's program to check what's happen.

What kind of system do you have?

Andre

[nave]

We're using Dell M610s. Is there other info you need?

[AndreTheGiant]

Try to install Dell Open Manage (version 6.1.0 works fine on vSphere) to see where is the problem.

Or check in the "DRAC" to see if there is some power-management feature that is active or some warnings.

Andre

[nave]

I checked the DRAC, and it seems OK. We have email alerts set up in the DRAC itself which have not triggered, so I think it's a VMware reporting issue, not actually a hardware problem. I'm starting to suspect that this is related to some of the Nehalem compatibility issues that I saw in this thread:

http://communities.vmware.com/thread/211585

I think I will look into opening a support issue. Thanks for your assistance!

[AndreTheGiant]

Have you also checked your hardware power management settings?

Andre

[nave]

I just went through everything again. The systems haven't logged any hardware issues in the last 3 weeks and we're not seeing anything with identical hardware (in the same blade chassis) that is not running VMware.

[azn2kew]

As mentioned, you can use vCenter to administer and monitor your resources and alarms with connections, CPU and memory nicely and set to send email notification if you want. If you have time spend some readings on vSphere Resource Management Guide here

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

Regards,

Stefan Nguyen

VMware vExpert 2009

iGeek Systems Inc.

VMware, Citrix, Microsoft Consultant

[azn2kew]

When you click on "Definition" button and select "New Alarm" and specify your percentage value on there so it doesn't have to be 75% or 90% as specified.

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

Regards,

Stefan Nguyen

VMware vExpert 2009

iGeek Systems Inc.

VMware, Citrix, Microsoft Consultant

[nave]

OK, I see how I could set up a new alarm that measures CPU status, but that still doesn't answer my original question, which I will rephrase slightly: What triggers the "host processor status" alarm? It's not strictly CPU usage, that's a different alarm ("Host CPU usage"). The host processor status alarm triggers on events (like power off or IP conflicts), not system status (like CPU/memory usage), so something else must be setting the hardware health status to red. What does that?

[jgowdy]

I too am experiencing this alarm issue with vSphere 4.0 and Dell M610

blades powered by Nehalem Xeons. I have six identical, brand new

blades in my new vSphere cluster, and they all trigger the alarm. The

alarm states that IERR# is being asserted on the second CPU (I have

dual quad core Nehalems).

[AndreTheGiant]

Have you tryied to install OMSA and see where is the problem?

Andre

[jgowdy]

With the M1000e blade chassis, the chassis interface serves as the management software. As OP said, it emails you when there's a problem with one of the blades that detectible by management. In this instance, it's an issue with vSphere itself. It seems the problem is a known issue in vSphere with large pages / TLS. See this thread: http://communities.vmware.com/thread/211585

[nave]

I set Mem.AllocGuestLargePage to 0, like was suggested in that thread, and all of my issues vanished. Reporting and alarms are down to what I hope are normal levels. Fortunately we don't need to use the large pages in our environment. Thanks for confirming, jgowdy.

If anyone does know what specifically causes the "host processor status alarm" to trigger under normal circumstances, I'm still interested. :smileyhappy: