I’m getting occasional failures when running on k8s that return “exit code None” at which point it doesn’t retry the job. Are there rules somewhere in the code about when a retry is attempted? I have the retry decorator on the task and it does successfully retry other errors for the same task (e.g. OOM)