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

How often does OneAgent report data to a metric?

Deekshacp
Dynatrace Participant
Dynatrace Participant

Hi Everyone, 

I have a use case where I need to increase the sensitivity of a metric event to create a problem card within seconds after it sees the anomaly. In the Advance Modal properties, I can set to 1 for violating samples and that is the highest I can set not less than that. This means problem will be created only if anomaly is seen for 1 minute continuously. So hence, I am wondering is this the maximum sensitivity we can set on is there any way we can reduce the same? 

 

Your ideas are appreciated.  

5 REPLIES 5

ChadTurner
DynaMight Legend
DynaMight Legend

This would be dependent on the metric. For example if it is a metric(s) from an extension/plugin that is only querying every 1,5,10,15 mins etc, then you are subject to that Query Frequency. 

If this is an Out Of The Box metric, then you are correct in the fact that the metric alerting sample will be at minimum 1 in the sliding window of up to 60 Mins. 

-Chad

Deekshacp
Dynatrace Participant
Dynatrace Participant

Thanks, Chad. That answered my question.

AntonioSousa
DynaMight Guru
DynaMight Guru

@Deekshacp@ChadTurner,

There has been some evolution recently. As can be seen in the configuration itself, for instance for CPU resource usage at the host level, it's not 1 minute anymore, but 10 seconds. Not well documented yet, but clearly visible:

AntonioSousa_0-1710962525697.png

 

Antonio Sousa

Hi @AntonioSousa 
This sampling duration happens only for the entity based anomaly detection settings not for metric events. For metric events, we still have the minimum of 1 min because builtin metrics values are reported to Dynatrace on every minute granularity. In services and user actions samples decreased to 30-seconds as well. 

Love more, hate less; Technology for all, together we grow.

That is exactly right, and even if the system is sampling the data at a 10-30 second frequency, the custom metric event will take that as an aggregate and then alert on it in a 1 min sampling. So say CPU, collected every 10 seconds.... 08:00:00 =20%, 08:00:10 =90%, 08:00:20 =10%, but if you are in the data explorer or using the metric selector, you can select the MAX value at the given time which in his case is 90% at 08:00:10, but you have to still wait for that 1 Min to elapse. 

-Chad

Featured Posts