Right now the retry backoff timing is hardcoded in the worker. We should have it be a globally configurable default proc, as well as configurable on a per job basis, as some jobs/conditions might warrant longer backoffs. Ideally the proc can take in the job stats (which including release count) and the job data/class, and return the number of seconds to delay
Right now the retry backoff timing is hardcoded in the worker. We should have it be a globally configurable default proc, as well as configurable on a per job basis, as some jobs/conditions might warrant longer backoffs. Ideally the proc can take in the job stats (which including release count) and the job data/class, and return the number of seconds to delay