Monitoring Event Queue

[kotterbein]

I am showing a large amount of schedueld events at the left end of the monitoring enging event queue, with basically nothing or otherwise flat at the right side of the spectrum, telling me that I'm overtaxing the computer per the Maximizing XI Performance doc. (averaging between 2688-3600 scheduled events).

I have Mod_Gearman in place, and the workers do not have any jobs waiting, and do not seem stressed at the least.

There is little to no latency in any check, the CPU profile is not high except for MySQL (averaging approximately 17% CPU Util), I do not seem to have any sort of execution time of host check latency showing at all. I'm just wondering if I should be concerned with flattening the Queue profile accross the 5 minute schedule, and if there is anything I can do to get to that type of profile other than what is in the doc currently. (most of which has been applied already).

any thoughts or help is appreciated!

[mguthrie]

Check to make sure that the time is synced on all servers? That can also throw off that chart.

If your latency would be high, then your bottleneck would probably be Disk IO, but that doesn't sound like the problem.

[mikew]

I am wondering if the change to monitoring based on seconds instead of minutes has an impact on the standard reporting features of Nagios XI.

This company is doing 10 second checks up to 10 minute checks.

Could that time range have an impact on the reporting?

[mguthrie]

It's possible. I don't think we've run any tests in-house on reports with the interval_length is changed, since it's a fairly unique situation where that is needed. Everything in Nagios is measured in unix timestamps, so most things should still calculate down to the second, but probably not fractions of seconds, which could make the monitoring event queue look very bunched up on the left side.