Unexpected behaviour on repeated job even if result_ttl > interval

[salander93]

As stated in the README:

IMPORTANT NOTE: If you set up a repeated job, you must ensure that you either do not set a result_ttl value or set one that is larger than the job interval. Otherwise, the job's details will expire, and the job will not be rescheduled.

However, I’m encountering a slightly different issue.

I have set my result_ttl as int(1.5 * interval)

Here is my setup (a semplified one):

• 1 worker
• 1 scheduled job (SJ_2m) with an interval of 2 minutes (the job takes 1 minute to complete)
• 1 scheduled job (SJ_60m) with an interval of 60 minutes (the job takes 10 minutes to complete)

Initially, I created SJ_2m, and it get scheduled and ran smoothly. Then, I added SJ_60m.

Since SJ_2m was created earlier and has already completed successfully, it is now in a SUCCESS state with a TTL of 2 minutes * 1.5 = 3 minutes (180 seconds).

When the next run time for SJ_2m arrives, it is added to the queue, but since the only worker is still busy (e.g., for 6 minutes), SJ_2m must wait. However, because the result_ttl is less than the time the worker becomes available (i.e., less than 6 minutes), the job details expire, and the job disappears, preventing it from being rescheduled.

I’m currently facing this issue in my project, and I believe the note should emphasize that even if you set a result_ttl value larger than the job interval, there is still no guarantee that the job details won’t expire. As a result, the job may not be rescheduled as expected if the worker is unavailable for longer than the result_ttl value.

[owlen]

I would also suggest that the IMPORTANT NOTE warning be implemented as a check in the code, and potentially throw an exception if result_ttl is set and not larger than the interval.

I was about to offer a PR for this, but seeing 77 open PRs pending, I'm not sure if this project is actively maintained. I would appreciate any input on this!