How is CPU saturation really calculated?

[AntonioSousa]

The documentation says:

By default, Dynatrace alerts when average CPU usage is higher than 95% in 3 of 5 consecutive one-minute intervals or when usage is higher than 90% in a single five-minute interval.

But, when we look at how we define custom settings for the CPU saturation settings, it seems to be based on the 10-second samples that are taken each minute:

Now, even though the two approaches might seem similar, there are cases where it will not happen. Given that we don't have visibility into the 10-second samples (we do have the min/avg/max values), how do we know what is exactly happening with CPU saturation detection?

BTW, remembers me of this Product Idea: https://community.dynatrace.com/t5/Product-ideas/RFE-Increase-resolution-of-data/idi-p/166920

[bradley_danyo]

Hi,

What documentation exactly are you looking at?  The documentation here discusses event thresholds using 10 second intervals.  I believe 10 second intervals are used for infra alerting, and one minute intervals are used for graphing.  If you provide me the documentation where it says that, I can get clarification from product team and get updated appropriately.

Regards,

Bradley

[AntonioSousa · Accepted solution]

@bradley_danyo,

The reference from documentation is here: https://www.dynatrace.com/support/help/platform/davis-ai/basics/events/event-types/resource-events

[bradley_danyo]

Thanks.  I've reached out to confirm and will let you know once I have more information.

Regards,

Bradley

[bradley_danyo]

Hi Antonio,

I've confirmed that the documentation you shared should be changed.  The CPU alerting now uses 10 second intervals.  Thank you for bringing this to our attention.

Regards,

Bradley