This repository was archived by the owner on Mar 20, 2023. It is now read-only.
This repository was archived by the owner on Mar 20, 2023. It is now read-only.
Kill old jobs on start #253
Description
At this moment, if Benefice gets restarted for whatever reason (config update, new version, etc), it will leave existing workloads running, and those will never get killed.
We should make sure Benefice kills all old jobs.
Some example ideas:
- Add a docker label with the calculated termination timestamp for a job (this would also ensure that everything gets killed the same way: we could get rid of the in-process timer).
- Store the fact that a container was started by Benefice, and on start, just kill all existing Benefice-started jobs.
- Possibly others...
I personally think the first idea might be best, as it would simplify code quite a bit possibly, and also avoids other bugs that'd avoid jobs not getting killed.
The actual killing of jobs could even become an external script that just gets executed by a systemd timer.