Nagios XI Check CPU alert levels

[almonitoradmin]

I have a question. I've run into an issue where I need to explain why we did not get any Nagios alerts when our linux MySQL server was running at about these levels for several hours:

user: 75%
system: 18.5%
iowait: 0.32%
idle: 6.02%

If I add the user and system together, it's roughly 93% cpu usage. I set our service check to be:

check_cpu_stats.sh -w 85 -c 95

So according to my reasoning, we should have gotten warning alerts.

When I check on the code that comes with the Nagios XI agent, it appears that it is measuring on IOWait. Is that something we can confirm?

Secondly, why would we want to alert on IOWait and not on a combination of SYSTEM and USER for a busy CPU? If those stats are very high, but IOWait is low, does that not create any problems? I would have though that the check would be measuring all of them and taking all of these into account.

Can anyone provide an explanation? We provide monthly graphs of our usage to one of our clients and the question is going to be "If the CPU was that busy, why didn't you get any alerts?"

[almonitoradmin]

I did notice going back into the Nagios XI Wizard it does specifically say:

CPU Statistics
Monitors the server CPU Statistics (% user, system, iowait, and idle)
I/O Wait Warning Level: I/O Wait Critical Load %

Why does Nagios XI alert on the IO Wait and not on the user and system stats?

[tmcdonald]

From where did you download check_cpu_stats.sh? That is not one of our standard plugins.