Send failed JobSink events into DLQ? #8779
Description
Problem
JobSinks are used for long-running jobs/tasks and are nicely embedded in the Knative-eventing. It "simply" creates a K8s Job per Event and mounts it into /etc/jobsink-event/event.
We use it since almost a year and we really like that feature!
Here and there our Dev's introduce some programmatic mistakes. The K8s Job recreates a failing Pod automatically - so like a retry-mechanism out of the box.
However, in some cases (programmatic mistakes) retrying doesn't help and it should end up being in the DLQ. It seems the JobSink approach doesn't have that feature...
Is it possible to add some logic into the JobSink controller to send the event into the DLQ if the job completely failed?
In my opinion that is the final feature to be fully embedded in the Knative-eventing world =)