How to configure monitor on sustained CPU or memory level

[w-bao]

Is there a way to configure CPU or memory monitors to send alerts only when the usage above the threshold is sustained for a period of time? e.g. If Memory warning threshold is 95%, once Nagios detects it reaches above 95%, it will not send out a warning unless that situation lasts for longer than say 3 minutes or even longer. Is there a way to do that?

[mcapra]

Yes in terms of when notifications get dispatched. You would do that in the "Check Settings" using the "Retry Interval" and "Max Check Attempts settings":

2017_02_27_09_55_49_CCM_Nagios_XI.png

Essentially, the "Retry Interval" is what kicks in and determines how often Nagios XI is checking a service when a problem is first detected. The "Max Check Attempts" them follows the retry interval to determine how many "soft states" or problems are detected before action is taken (notifications, event handlers, etc).

If you only wanted to receive notifications when a problem has been persistent for 3 minutes, you could set your retry interval to 1 minute and you max check attempts to 3 attempts. This would mean that, when a problem is first detected, Nagios XI will check every 1 minute up to 3 times before taking action.

If, for example, you only wanted to receive notifications when a problem has occurred for 10 or more minutes, you could set the retry interval to 1 minute and the max check attempts to 10 attempts. Or, you could set the retry interval to 2 minutes and the max check attempts to 5 attempts. In these cases, 1*10=10 and 2*5=10 though in the first case we are checking the service more aggressively.