+1
Planned

Load average checks

Jeff Uberstine 11 years ago updated by Konstantin (Here to help) 11 years ago 3

I would like to have a check for the local agents to check for CPU load average. It is not uncommon for CPU spikes on Virtualized machines. Very often a CPU may spike from 12% to 100%, then back down. In this case, an alert is thrown. I would like a check for load average. For example, if the load average is over 1.0 or 4.0 then throw an alert. I would like to set the load level at which the alert is thrown. This provides a much better indication of a problem and prevent false positives. Could this feature be added?

Answer

Answer
Planned

Ok, so we are talking about the same thing. Yes we have that in plans. Will keep you posted and thanks for reporting )

Just to doublecheck we are on the same page. 

Instead of CPU utilisation Monitor you'd like to have CPU load Monitor as described here http://en.wikipedia.org/wiki/Load_(computing) ? I assume that you see averaginng period same as/close to the Monitor period?

That is correct. For our needs, CPU load would indicate issues much more accurately then percentage. CPU spikes are not uncommon and generally do not cause issues in web server environments however a high Load average almost always means a problem is occurring or about to occur. I think both options, CPU load and CPU percentage, should be available.

Here is another good article that breaks it down in a much eaiser to understand way.


http://blog.scoutapp.com/articles/2009/07/31/understanding-load-averages

Answer
Planned

Ok, so we are talking about the same thing. Yes we have that in plans. Will keep you posted and thanks for reporting )