Autoscaling of runners doesn't seem to work properly

[rahul9170]

Hello,

I am trying to setup the controller v0.23.0 and facing issue with auto scaling have below args passed to controller deployment.yaml

RunnderDeployment.yaml (Replicas=0 has been set)

HorizontalRunnerAutoscaler.yaml (status.desiredReplicas=0)

After a successful automatic scale up of 1 runner, it should scale down as per scaleDownDelaySecondsAfterScaleOut setting but the runner pod continues to run. The controller logs also suggest the same but pod continues to run

Can you please help with this issue?

[rahul9170]

Also I get below entry in the logs

[toast-gear]

You've not really provided enough inforamation for anyone to help you, you don't need to hide log lines. The controller logs reasons why it isn't scaling something down. What additional log information is there that relates to scaling down decisions?

[mumoshu]

it should scale down as per scaleDownDelaySecondsAfterScaleOut

This perception can be problematic. You have --sync-period=1m so ARC should only able to sync the desired replicas according to the content returned by GitHub API approx once per 1m.

controller deployment.yaml

Also, how did you deploy ARC?
Did you really deployed ARC using kubectl or kustomize without helm even though you seem to be using helm for deploying runners?