How to ensure that overlapping jobs function properly?

[junoriosity]

In order outline my issue, I will try to make it as simple as possible. I am not very experienced with that library, so I hope you forgive me for my lack precision here.

• I have a FastAPI server running and want to use a scheduler (essentially I want to follow this one here).
• I have a function I want to run in the scheduler
• I expect that function to run 500 ms.
• Now I would like to run that twice. Job 1 like now, then in 500 ms, 1000 ms, 1500 ms, ... Job 2 like now + 250 ms, then now + 750 ms, now + 1250 ms. That way we would run it overlapping, and the processing latency (until start) would drop to 250 ms.

My questions are as follows:

• How would I implement in the best way?
• What happens if a job, say job 2, is stuck?
• What happens if the job takes longer than 500 ms? Is then a new one triggered and we run the Job 2 twice or thrice in parallel? Or is it stopped then and replaced? Or does one wait somehow?

It would be great to some input from you here. 🙂