cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

This product reached the end of support date on March 31, 2021.

Application Process Unhealthy - Short Evaluation Timeframe

miguel_balsa
Dynatrace Mentor
Dynatrace Mentor

Due to the short evaluation timeframe of this incident, there are too many alerts raised in a short period of time.. this should be only one alert.

Also, in the description it says "in the last 5 minutes", which conflicts with the Incident Rule configuration, that says 10 seconds.

Is there any way to avoid this amount on alerts for an incident that is obviously just one? It only happens on OOTB Incidents.

Thank you!! 🙂

4 REPLIES 4

christoph_wiesi
Dynatrace Helper
Dynatrace Helper

Hi Miguel,

you can ignore the evalution timeframe of 10 seconds. Internally we use 5 minutes as the description correctly states.

With the public updates 6.3.27 and 6.5.17 we changed the calculation of this incident a little bit to avoid possible false/positives, like in this case.

The GC health calculation with the update is now the following:

  • every minute we aggregate the sum of the gc suspension time
  • then we check the aggregated value against a fixed time (15 % of a minute)
  • is the aggregated value 4 out of 5 times above the threshold we trigger an incident

Best Regards,

Christoph

Hi @Christoph W., thanks for the answer, that was very clear.

Although, I still have two doubts:

- If the evaluation frame is 5 minutes, how do I have incidents that last only 54 seconds?

- Is there any way to change/adapt this behavior? There has been 5 incidents raised in 15 minutes for the same problem in the same server, which is kind of annoying for the customer, being clearly the same issue/problem.

Thank you!

Miguel

Hi,

1. Because we run a cyclic task in the background, which checks every minute the aggregated value against the thresholds. Maybe the image under evaluation timeframe https://www.dynatrace.com/support/doc/appmon/administration/system-profile/edit-a-system-profile/system-profile-incidents/configure-incident-rules/#&gid=0&pid=1
can illustrate this better.

2. If you haven't installed the latest update you can change the thresholds by yourself (also in the client debug ui) with the following debug flags: (server restart is necessary)
https://www.dynatrace.com/support/doc/appmon/infrastructure-monitoring/process-health-monitoring/#process-health-incidents

Best Regards,

Christoph

Hi Christoph,

Is this change the same of many of the Out of the Box alerts like the application process disconnected alert? that alert is in the same boat where it shows a 10 second evaluation and that is unchangeable.

thanks,

Rob