Starting existing container fails

[dpwrussell]

If you create a container in the usual way, stop it and restart, the following warning occurs. I think we should work around that for the docker case as starting a container you know there can not be existing running processes.

web_1    | 0 static files copied to '/opt/omero/web/OMERO.web/lib/python/omeroweb/static', 611 unmodified, 2 post-processed.
web_1    | Clearing expired sessions. This may take some time... [OK]
web_1    | Starting OMERO.web... [ERROR] OMERO.web workers (PID 79) - no such process. Use `ps aux | grep /opt/omero/web/OMERO.web/var/django.pid` and kill stale processes by hand.
web_1    | Removed stale /opt/omero/web/OMERO.web/var/django.pid

[manics]

Sounds like a bug in OMERO.web- if it's terminated with docker stop (not docker kill) it should shut-down cleanly and remove it's PID file. Might need a signal handler added to it?

[dpwrussell]

I just confirmed that docker-compose is using docker stop as opposed to docker kill and it certainly is. I also tried start -> stop -> start with docker, without compose and the error is the same.

[manics]

If I run omero web --foreground locally (not in container) Ctrl-C causes a clean shutdown:

$ OMERO.web/bin/omero web start --foreground
Post-processed 'omeroweb.viewer.min.css' as 'omeroweb.viewer.min.css'
Post-processed 'omeroweb.viewer.min.js' as 'omeroweb.viewer.min.js'

0 static files copied to '/home/spli/OMERO.web/lib/python/omeroweb/static', 611 unmodified, 2 post-processed.
Clearing expired sessions. This may take some time... [OK]
Starting OMERO.web... [2017-07-25 15:15:17 +0000] [14389] [INFO] Starting gunicorn 19.7.1
[2017-07-25 15:15:17 +0000] [14389] [INFO] Listening at: http://127.0.0.1:14080 (14389)
[2017-07-25 15:15:17 +0000] [14389] [INFO] Using worker: sync
[2017-07-25 15:15:17 +0000] [14394] [INFO] Booting worker with pid: 14394
[2017-07-25 15:15:17 +0000] [14395] [INFO] Booting worker with pid: 14395
[2017-07-25 15:15:17 +0000] [14404] [INFO] Booting worker with pid: 14404
[2017-07-25 15:15:17 +0000] [14405] [INFO] Booting worker with pid: 14405
[2017-07-25 15:15:17 +0000] [14414] [INFO] Booting worker with pid: 14414

^CCancelled
[2017-07-25 15:15:29 +0000] [14389] [INFO] Handling signal: int
[2017-07-25 16:15:29 +0000] [14394] [INFO] Worker exiting (pid: 14394)
[2017-07-25 16:15:29 +0000] [14414] [INFO] Worker exiting (pid: 14414)
[2017-07-25 16:15:29 +0000] [14404] [INFO] Worker exiting (pid: 14404)
[2017-07-25 16:15:29 +0000] [14395] [INFO] Worker exiting (pid: 14395)
[2017-07-25 16:15:29 +0000] [14405] [INFO] Worker exiting (pid: 14405)
[2017-07-25 15:15:29 +0000] [14389] [INFO] Shutting down: Master

docker stop doesn't do a clean shutdown, defining STOPSIGNAL SIGINT doesn't seem to work either. This requires more investigation.

[joshmoore]

Is an exec perhaps needed for the last file?

https://github.com/Yelp/dumb-init#using-a-shell-for-pre-start-hooks

The call hierarchy is likely:

• docker
  • dumb-init
    • bash
      • 99
        • exec python

which would get rid of the 99 shell but not the for-loop one.

[manics]

I tried that before, and also tried adding a signal handler without success.

[manics]

I've opened a PR with the suggested workaround #14

This issue should remain open until we figure out why OMERO.web isn't being shutdown cleanly.