@catte i know right? The best I understand is that a failed job is one that is retried whether or not it ultimately succeeds. It would be nice to have initial fails write to a log though