Host Power supply alert is not going off

[esxi1979]

I had a Host which had bad power supply, I got alert in vcops. I replaced the same. But now 10 days over the alert is not going off

Any clue ? I see an option to cancel the alert but I am sure it should go green of its own

[gradinka]

If you're positive the problem is resolved, go ahead and cancel it.

It is possible that vcops didn't get the "cancellation" event from VC on that, for reasons unknown - it happens, especially in cases of hardware-related faults (storages down, faulty PS, etc).

So most likely it won't go away on its own.

[mark_j]

If it is a Fault, it will not clear itself. Faults need to be cleared manually since they're initiated by "events" rather than "conditions", so there is no logic to tell vC Ops that the event is over. E.g. vSphere will fire an event if a disk failed, but it doesn't necessarily fire another event when the conditions fixes itself.

Sometimes it's pulling the data from vCenter, and vCenter is getting data from the vSphere, but the vSphere keeps getting the alert from the motherboard. Sometimes when you have these types of errors, you need to go in and clear the SEL/ motherboard logs so they stop reporting.

[gradinka · Accepted solution]

mark.j wrote:

If it is a Fault, it will not clear itself. Faults need to be cleared manually since they're initiated by "events" rather than "conditions", so there is no logic to tell vC Ops that the event is over. E.g. vSphere will fire an event if a disk failed, but it doesn't necessarily fire another event when the conditions fixes itself.

This is partially true - there is logic in vcops which will wait for an "opposite" event, and cancel the fault if that is received.

But (as you point out) not all events have exact "opposite" or "closing" events;

or, that opposite event might have some different metadata associated with it, so it's not possible to make that 1:1 mapping automatically and cancel the fault

[esxi1979]

I agree with "But (as you point out) not all events have exact "opposite" or "closing" events;" & other statements as well, as I had "fault" in vcenter for some storage related service, it went off as i started the service, but the h/w fault i cleared manually

thanks folks