Closed RothAndrew closed 1 month ago
Was the status of these failed
? I don't have failed workflows set up to GC, so we could do one of the following:
Change controller.workflowDefaults.spec.ttlStrategy
to secondsAfterCompletion
instead of secondsAfterSuccess
. OR, we could set a controller.workflowDefaults.spec.activeDeadlineSeconds
which would limit the amount of time a workflow can run regardless of status (I'd probably set it to something like 3600).
The status was Running
, but the actual pods were still 0/3 since Kubernetes wouldn't start them due to the missing secret.
OK then we probably need a activeDeadlineSeconds default param I'll throw one in and PR it
Yesterday I submitted some workflows and ran in to #32. This morning I saw they were still in the cluster still trying to start. We should figure out how to set a timeout so that the argo controller kills them if they aren't finished within a certain time period.