As a side note, taskotron01.qa is currently low on disk and triggering nagios warnings. Since that's a production system, the free space isn't critical yet and infra freeze lifts tomorrow, we'd deal with it after freeze to avoid any freeze break requests

I haven't been able to find out why it's not working after an hour or so debugging. However, I used tmpfiles.d to clean the directories on taskotron-dev, if that works, I'll put that into ansible and deploy to our taskotron masters if there are no objections.

For some reason, even tmpfiles.d doesn't work on /srv/taskotron and /srv/buildmaster. It does work though on /var/lib/taskotron-trigger/cache. Since I don't know why tmpwatch nor tmpfiles.d work on those dirs, I set up cronjob that cleans the dirs with find, let's see if that works.

I set up cronjob that cleans the dirs with find, let's see if that works.

This seems to work, it's in ansible for dev and stg. I'll put that for prod tomorrow, if there no issues. Also, I think that 4 months old logs take up 80% of disk, so we might be getting nagios warnings either way. We might want to decrease to 3 months until we rebuild and have more disk space.

OK, seems to be finally working, I am going to close this. Please reopen if any issue appears.