Complex Metrics

[krobertson71]

Is there any way to create complex metrics?

Bascially the theory is: I don't care if the CPU is at 98 percent as long as the CPU Queue is not above a 10.

I haven't noticed any way to achieve this, but I figured if there is, this is the place to ask!

[sreinhardt]

With individual checks in most cases, this isn't possible at the plugin level. However you could write a wrapper script to do this. Additionally, service dependencies could achieve this to some extent, in that you could make the load dependent on the percentage check, and never alert on percentage. Then you would only get alerts when percentage is high and load is high. However this might cause issues as you might also have times where load might be high, but percentage is not, unlikely but something that might happen and be better handled through scripting.

[krobertson71]

Yeah the service check dependency did cross my mind but it , too me, makes the cpu % check of no use as you could just alert on the load and forget the percentage or am I looking at this the wrong way?

[tmcdonald]

Almost sounds like BPI might come in handy here:

http://assets.nagios.com/downloads/nagi ... BPI_v2.pdf

It takes some getting used to, but does increase the complexity you are able to achieve.