On off2, Apache not restarted after oom kill, even with Restart=on-abort

[stephanegigandet]

On Friday and Saturday the Apache server on off2 for off (container 113) got killed by oom, and the service was not restarted automatically even though we have "Restart=on-abort" in off:/etc/systemd/system/multi-user.target.wants/apache2.service

We should find a way to prevent the oom, and we should make sure that Apache restarts if it gets one anyway.

https://openfoodfacts.slack.com/archives/C1FPYCWM7/p1705046870490599?thread_ts=1705046654.492669&cid=C1FPYCWM7

stephane
  [il y a 3 jours](https://openfoodfacts.slack.com/archives/C1FPYCWM7/p1705046870490599?thread_ts=1705046654.492669&cid=C1FPYCWM7)
[Fri Jan 12 01:00:57.396486 2024] [mpm_prefork:notice] [pid 2455110] AH00169: caught SIGTERM, shutting down

stephane
  [il y a 3 jours](https://openfoodfacts.slack.com/archives/C1FPYCWM7/p1705047049772889?thread_ts=1705046654.492669&cid=C1FPYCWM7)
apache for off was stopped, restarted

stephane
  [il y a 3 jours](https://openfoodfacts.slack.com/archives/C1FPYCWM7/p1705047191830349?thread_ts=1705046654.492669&cid=C1FPYCWM7)
Jan 12 01:00:52 off systemd[1]: apache2.service: A process of this unit has been killed by
the OOM killer.
Jan 12 01:01:03 off systemd[1]: apache2.service: Failed with result 'oom-kill'.
Jan 12 01:01:03 off systemd[1]: apache2.service: Triggering OnFailure= dependencies.
Jan 12 01:01:03 off systemd[1]: apache2.service: Consumed 2w 5d 10h 7min 22.241s CPU time.
Jan 12 01:01:03 off systemd[1]: Starting apache2-off failure email notification...
Jan 12 01:01:04 off bash[2729420]: Unit apache2-off.service could not be found.
Jan 12 01:01:05 off systemd[1]: email-failures@apache2-off.service: Succeeded.
Jan 12 01:01:05 off postfix/pickup[2727098]: 150F718EFD6: uid=0 from=<root>
Jan 12 01:01:05 off bash[2729421]: Null message body; hope that's ok
Jan 12 01:01:05 off systemd[1]: Finished apache2-off failure email notification.

It's strange that Apache is logging a SIGTERM