I need to create an incident rule based on a mesure to detect a no-activity period (for example : 1 hour without any purepath).
Is it possible to do that with AppMon 7.2 ?
If I create a measure (web requests count) with a lower severe threshold to 0 (zero), it seems the incident is never raised.
Thanks four your help
For these types of situations first I like to graph a measure over a period of time and then look at it relative to the aggregation I'm interested in. For your example above for instance I would chart web requests, set the aggregation to minimum and then look at when and where it passes that lower threshold you are concerned about. Then you can setup the limits and create the incident rules based on that.
I've created a number of these Incident rules that work as expected (on 6.5 and 7.1).
Is this something that has stopped when you upgraded to 7.2?
If not then just to make certain:
Have you created the measure on a metric that does have a count value (Not an implied count value)?
On the Conditions tab of the Incident Rule have you set the Aggregation to 'count'?
Have you made a chart for your measure that proves you are receiving zero transactions?
Thanks a lot for your reply.
No, it didn't stop working upon migration as this rule has been created with 7.2 only.
The measure is a Web Services Count with criteria on methode name, and lower severe threshold = 0
The Aggregation in conditions tab is set to "sum" : I change it to "count" to see the difference.
And yes, I have a chart showing no activity during a long period (night hours in my case).When I set the condition to threshold "warning or severe" (having a lower warning = 10), the incident is created when less than 10 purepaths are measured, but not at 0 purepath
Ok a couple of other things to check/try.
Is the evaluation period of the Incident Rule short enough for a zero transactions to occur throughout that period?
Can you also try and create another measure for your Web Service method. Create a Method>Invocation Measure. Ensure that it's capturing correctly with either Sum or Count aggregation. Then set the lower thresholds both warning and severe to 0.
Create another Inc Rule for it with a valid evaluation period and set the measure to fire on Warning or severe.
See if this give a different result.
The evaluation period is currently 15 min, and I have many hours of inactivity (from 21pm to 7am), so the incident should raise.
I try to create a new measure and a new incident rule... and I will let you know.
The result is successful : I now have incidents based on 0 activity. 🙂
I dont't know exactly if the problem is on the the condition aggregation (count vs. sum) or on the measure itself. I will carry on further tests to have a correct idea.
Thank you very much for your help.