Replies: 1 comment 3 replies
-
@jgreat Note that the duration starts when ARC added the replica, not when the runner started running a workflow job. Are you sure your cluster is responsive enough so that it can create a runner pod and it finishes running the job within 1m after github-webhook-based autoscaler of ARC received the webhook event for a queued job, and that works universally even when you triggered several workflow jobs concurrently? (I think it's not. |
Beta Was this translation helpful? Give feedback.
-
Our runs get stuck sometime because the controller isn't back filling pods after completion. I'm sure I have some settings wrong, but can't figure out why the controller delays reconciliation and refills the pods. Any insights would be amazing.
Helm values.yaml
manifests for "small" runners
Controller logs:
Beta Was this translation helpful? Give feedback.
All reactions