Process group availability

Platform Dashboarding
1

How can I create a metric on data explorer for process group availabilty spliting by process group?

2

hello @kobkob you can use builtin:pgi.availability

DanielS_0-1714487301027.png
DanielS_0-1714487301027.png2128Ă—126 20.2 KB

3

Hi @DanielS thanks for your response.

I used it, but its not possible to split it by process group, but only by process.

In the end I want to monitor the cluster. For example if I have a service with 2 instances and 1 of them is down, I want to see 50% availability.

Thanks

4

Understood, do you have enabled process group availability monitoring?

5

I leave you the link Process group availability monitoring and alerting — Dynatrace Docs

6

Yes, it’s enabled.

7

This is not correct.

The process is not available but the process Group is available because the second one is handling the load.

I suggest you to split the Graph by process (and filter if needed), with this way you can know if the process i with 0% availability or not.

Malaik_0-1714558331286.png
Malaik_0-1714558331286.png953Ă—284 9.42 KB

8

Hi @kobkob 
I think its a great idea of Malaik  and DanielS

PierreGutierrez_1-1715267926360.png
PierreGutierrez_1-1715267926360.png1453Ă—706 19.3 KB

Alternative 1 : in that way you can use this Metric Selector

builtin:pgi.availability:filter(and(or(in("dt.entity.process_group_instance",entitySelector("type(process_group_instance),fromRelationship.isInstanceOf(type(PROCESS_GROUP),entityName.equals(~"bin/npm-cli.js (npm) ms-insurance-bs-*~"))"))))):splitBy("dt.entity.process_group_instance"):sort(value(auto,descending)):limit(20)

I have found another metric that can be much more useful for you, however it is based on the count of instances (Not exactly on their availability %). In my personal opinion I think this metric is more useful, the reason is because the “Process group availability monitoring” rules are also based on the number of Processes or process instances to alert.

The great thing about this metric is that it has the Splt by “Process Group”.

Alternative 2 :   Use the metric “Process count per process group”   ( builtin:tech.generic.count )

PierreGutierrez_2-1715268649905.png
PierreGutierrez_2-1715268649905.png1456Ă—759 25.7 KB

PierreGutierrez_3-1715268714993.png
PierreGutierrez_3-1715268714993.png1018Ă—664 23.2 KB

builtin:tech.generic.count:filter(and(or(in("dt.entity.process_group",entitySelector("type(process_group),entityName.equals(~"ms-bff-crm~")"))))):splitBy("dt.entity.process_group"):sort(value(auto,descending)):limit(20)

I’m sure this will be very helpful :flexed_biceps::blush:

9

Additionally if you want process group availability metric without explicitly specifying the group (for example for a Management zone):
For single value charts in Data Explorer such as table:

(100*(builtin:pgi.availability.state:filter(eq("availability.state",available)):splitBy("dt.entity.process_group"):count 

/ builtin:pgi.availability.state:filter(eq("availability.state",available)):splitBy("dt.entity.process_group"):count) 

:default(0)):fold(avg):setUnit(Percent)

For charts just remove the :fold(avg) Using the fold aggregation in Data Explorer won’t work properly due to way how Dynatrace queries single value charts (infinite resolution), so use :fold in such cases.