Avg CPU Load for Cisco Catalyst 37xx switch wrong; shows 99% and switch says less than 10%

PHA

I have a Cisco Catalyst 3750 at one of my sites. It is showing Avg CPU Load of 99% on Node Details (and also on the alert I have emailed to me, when CPU goes over 90%). The weird thing? If we telnet directly to that switch and have it show utilization (I'm not the switch guy, so I didn't enter the commands, I only saw the result), it is showing utilization under 10%. Now, it did spike at 99% last night, but has since dropped. But I'm not seeing the drop in NPM. I have tried polling again, I have verified that I am using the SNMP community. But the Node Details stills shows 99%, and the alert keeps emailing the same thing, since it doesn't seem to be resetting ... but why isn't it resetting? My other 3750s aren't exhibiting this problem, just this one switch. And it seems to be NPM's problem, not the switch's.

So where should I go from here, to investigate further?

PHA

Never mind; I seem to have solved it .. .looks like I was wrong, and it was using the wrong SNMP community. Upon further consultation with my switch guy, we determined that he changed the SNMP community earlier, when I asked him to, and I changed the SNMP community on the discovery. BUT ... I didn't change it on the node details.

So it looks like it was reported the last thing it saw with the old community, and since it couldn't talk to it anymore with that community, it never was able to clear the counters.

When I re-entered the community name for this particular node, and re-discovered/re-polled, it dropped down to a more proper 6%.

So I'm all good.