Closed pcm32 closed 3 months ago
I am going to start work on this pronto as I just ran out of space on a 2TB volume. I will add as many of the jobs that main uses as I can.
That's great! It's also worth checking whether we can adapt something like this: https://github.com/usegalaxy-au/infrastructure/pull/1733/files#diff-cf003cea0be89303fa5922fc7be3947c14bb07c1d32add1e233381589e7be2aeR82
Instead of using the max wall time, it checks the oldest running job to determine what to clean.
I just saw your comment and we could definitely do something like that. We may also want to check the job status, but baby steps.
Currently the maintenance job https://github.com/galaxyproject/galaxy-helm/blob/master/galaxy/templates/cronjob-maintenance.yaml deletes the galaxy tmp. We should probably add there:
/usr/bin/tmpwatch -v --mtime --dirmtime 7d
on the jobs working directory (or at least on_cleared/
, which is the result of the resubmissions - the initial jobs that didn't work).