We are running a workflow on a cron schedule of 2 mins. We want the pods to not linger around for long after completion. I think I saw somewhere that the default GC time is 30 mins. Is there a way to reduce this time? We tried

delete-resource-on-finalize

which kills the pods immediately and we are losing logs due to this as Datadog is not able to that quickly scrape logs from short lived pods. I see there is a way to do this via configmap, we are using

flyte-binary

deployment, I can't find where exactly to put this setting? Can someone help?

Have you tried setting

gc-interval

in the admin configuration?

Yes, I tried putting it in two places in the

flyte-binary

values.yaml but it didn't work.

configuration:
  inlineConfigMap: """
    schedulerConfig:
      gc-interval: 2m #can leave blank for 30m
      delete-resource-on-finalize: false
      """

and

flyte-core-components:
  admin:
    schedulerConfig:
      gc-interval: 2m #can leave blank for 30m
      delete-resource-on-finalize: false

Are either of these correct? Or is there some other section I should be putting this setting under?

This has to work. Can you try setting it under propeller?