condition must exist for

Question

I may be over thinking this...

how does condition must exist for relate to evaluation frequency of an advanced alert?

Seems to me that the evaluation frequency is how often the alert "polls" for the condition while the 'condition must exist for' is how many polls the condition must exist for the advanced alert to trigger.

If that is in fact the case than why is the 'condition must exist for' drop down predefined for seconds, minutes or hours?

Seems to be the better option would be, set your evaluation frequency to X seconds, minutes or hours.

Then you set your 'condition must exist for' to "condition must exist for X evaluation frequencies"

I'll give you an example:

Say I have an advanced alert to monitor for high CPU and I want to evaluate the condition every 15 minutes, so 4 times an hour.  However I only want the advanced alert to trigger if the high CPU condition has persisted for 2 frequencies.

The current implementation of these settings seems obvious for the evaluation frequency but not so much (to me anyway) for the 'condition must exist for'.  In my example, the evaluation frequency would be set to 15 minutes while the 'condition must exist for' setting would be set at 30 minutes?  Like I said, I'm probably over thinking this...

mikeit83 · Answer

You have to keep in mind that "must exist for" logic has to be grounded in polling intervals.    If you poll a device every 10 minutes-  its a good idea to keep your must exist for as a multiple of 10.

I usually have really sensitive mission critical polling done every 2 or 3 minutes.  So I can get an alert if its down for 6 minutes or more.

Just a thought...

happyfunnorm · Answer

Since it evaluates every 15 minutes, to get 2 polling periods you only really need 16 minutes to be sure to get 2 polling periods.

mrxinu · Answer

I think if it were, say, a volume alert and you knew that 100% of your volumes polled at the exact same interval, then having a setting for the # of intervals would make sense. But since you can create an alert for volume consumption and the volume statistic sampling can differ from one volume to the next (even though the alert would cover all of them), that would cause some odd behavior.