Prometheus queries (queued jobs) for Actions Metrics Server

[duhow]

I am writing some Prometheus queries to create Grafana dashboards, but I'm losing my head with the incoming data...

Apparently, there is the case when (for example):

• There are 40 GHA jobs currently running
• Actions Metrics Server starts
• Metrics add +40 jobs completed
• There are 0 jobs started as the webhook call was not received/processed.

When running the following query, this causes that "base line" is over 0.

sum(sum_over_time(github_workflow_jobs_queued_total{runs_on=~"(.+)?runner-group(.+)?"}[60s])) - 
sum(sum_over_time(github_workflow_jobs_started_total{runs_on=~"(.+)?runner-group(.+)?"}[60s]))

In order to fix this, I have to manually substract a variable amount (330 in this case) to fix this visualization:

clamp_min(
    sum(sum_over_time(github_workflow_jobs_queued_total{runs_on=~"(.+)?runner-group(.+)?"}[60s])) - 
    sum(sum_over_time(github_workflow_jobs_started_total{runs_on=~"(.+)?runner-group(.+)?"}[60s])) - 330
, 0)

Any idea on how to fix this numbers or get the minimum value over time to substract automatically?

[Samsanga]

@duhow , Did you manage to get actions-metrics-server working on GitHub Enterprise or GitHub.com ?

[duhow]

Yes, we are using GitHub Cloud (GitHub Team).