windows machines and cpu alarms

[whoolly]

On occasion Nagios will report that a windows server has been at 100 % cpu. However, if you check the machine this is not the case, if you watch the machine with perfmon, it definitely does NOT have a 5 minute avg of 100 %.
This has happened with many different machines over the years.
In general, either restarting the NSCLIENT++ service on the Windows server, restarting the Nagios service on the server, or rebooting the Nagios server has cleared this up.
However, I now have one machine that is always reporting in Nagios that it is at 100% cpu and I can't get it to clear.

We are using Nagios Core 3.2.3 ... I know its old, but has always stood us well outside of these minor annoyances.

Anyone have an idea what causes this counter to get "stuck" or how to clear it?

I have tried resetting the performance counters on the Windows server, and perfmon has no issues.

Thanks in advance.

[rkennedy]

Can you post the check_command that is running for the service specifically? It could be alerting based off of only 1 core, vs the entire system, but it really depends on the check you're running.

[whoolly]

I use the same check on all of my machines ..
check_command check_nt!CPULOAD!-l 5,80,90

This particular server is a virtual with only 1 virtual cpu.

[mcapra]

We are using Nagios Core 3.2.3 ... I know its old, but has always stood us well outside of these minor annoyances.

It's more likely that there is an issue with the remote machine's NSClient++ in this case. Can you share which version of NSClient++ is being leveraged as well as the Windows version of the machine being monitored?