Is it possible to apply threshold to Process CPU based on the usage we analyzed

[Athira]

I want to know is there any feature in Dynatrace that allows to apply threshold settings for CPU consumption for a  Process. 

[natanael_mendes]

You can do this on the host level

Take a look on the documentation page

https://www.dynatrace.com/support/help/platform-modules/infrastructure-monitoring/hosts/configuratio...

go to host>settings>anomaly detection

[ChadTurner · Accepted solution]

If you are looking to impose an alert criteria on a process that runs on a host and alert on the process consumption outside of the host overall consumption, it can be done via Custom Metric Events: 

go to settings>anomaly detection>Metric Events - and add one in 🙂 

[Athira]

hi Chad, i have setup'd Custom Metric Events, to generate alerts for generic.cpu.totalTime , but i'm observing the alerts are raised for the The Process total CPU time value of 0 µs.

i have applied Static threshold of 14Mins, still the alert are raised for The Process total CPU time value of 0 µs was above your custom threshold of 14 min.

Can you advise, what settings or attributes can be configured to alert only when the CPU time cross the threshold

[ChadTurner]

Can you share a screen shot of the metric event you created? It should look something like this: 

[Athira]

please see the snaps attached

[ChadTurner]

Sorry for the delay, been pretty busy today. The reason why you are being alerted even at the 0 Upsidedow "h"s is because you have selected to be alerted on missing data. I suspect that there is occasionally a delay in the metrics being passed for example its looking for data every 1 min, 8:00 data arrives, 8:01 no data, 8:02 no data - At this point an alert is triggered, 8:03 current data and missing data from 8:01 and 8:02 shows up, giving you the illusion there was no gap in data but there was. 

Turn off the alert on missing data and see if your problems go away.