I'm running a Gunicorn server inside a schroot session via supervisor. My problem is that the service is not fully stopped when stopping it with "supervisorctl stop".
This is the script (simplified) controlling my server, it runs gunicorn in foreground:
# gunicorn.sh
schroot -c gunicorn -r -- bash -c "gunicorn --workers=1 myapp.wsgi:application"
This is my supervisor config to run this script:
[program:gunicorn]
command=/home/test/gunicorn.sh
stderr_logfile=/var/log/gunicorn.err.log
stdout_logfile=/var/log/gunicorn.out.log
When I start the service via "supervisorctl start" , my process tree looks like this:
supervisord(7175)---gunicorn.sh(8061)---schroot(8067)---gunicorn(8068)---gunicorn(8073)---{gunicorn}(8078)
Now when I stop the service with "supervisorctl stop", The corresponding supervisor process and its direct child, gunicorn.sh, are terminated. But the schroot process itself continues to live and is now a child of the init process:
schroot(8067)---gunicorn(8068)---gunicorn(8073)---{gunicorn}(8078)
This whole behavior seems to be related to both the way schroot and gunicorn work.
How can I let supervisor stop my schroot-hosted process correctly?
Had to start gunicorn with exec:
# gunicorn.sh
exec schroot -c gunicorn -r -- bash -c "gunicorn --workers=1 myapp.wsgi:application"