How can I run django-q qcluster with supervisor process manager

[GabLeRoux]

Hey there,
I like how simple it was to configure django-q :). I'm using supervisor to auto start the cluster daemon a bit like #102 . After a while (after a few deployments), I realized I had all my server memory taken by q-cluster processes. Our provisionner and deployment procedures restarts supervisor commands which explains the problem.

When I restart supervisor django-q command, it doesn't stop the cluster and instead starts a new process leaving orphan qcluster processes in the wild.

My supervisor configuration is quite simple:

/etc/supervisor/conf.d/django-q.conf

[program:django-q]
command = env $(cat /srv/project/.env | xargs) /usr/local/virtualenvs/project/bin/python /srv/project/manage.py qcluster
user = vagrant

I've read the following documentation:
http://django-q.readthedocs.io/en/latest/cluster.html#process-managers

There's an example config for circus which is quite simple, but we're already using supervisor for some other processes.

By default, superviosr's numprocs is set to 1 so I don't think this is the problem.
I've found the following related ServerFault question:
http://serverfault.com/questions/608069/managing-daemons-with-supervisor-no-foreground-mode-available

I tried it, but the problem still happens with the following config:

/etc/supervisor/conf.d/django-q.conf

[program:django-q]
command = /srv/project-django-q-runner.sh
user = vagrant

/srv/project-django-q-runner.sh

#!/usr/bin/env bash
set -eu

pidfile="/var/run/project-django-q.pid"
command="env $(cat /srv/project/.env | xargs) /usr/local/virtualenvs/project/bin/python /srv/project/manage.py qcluster"

# Proxy signals
function kill_app(){
    kill $(cat $pidfile)
    exit 0 # exit okay
}
trap "kill_app" SIGINT SIGTERM

# Launch daemon
$command
sleep 2

# Loop while the pidfile and the process exist
while [ -f $pidfile ] && kill -0 $(cat $pidfile) ; do
    sleep 0.5
done
exit 1000 # exit unexpected

Here's my settings.py's Q_CLUSTER just in case:

Q_CLUSTER = {
    'name': 'project',
    'workers': 1,
    'retry': 900,
    'queue_limit': 2,
    'bulk': 10,
    'orm': 'default',
}

I tried moving out the env $(cat /srv/project/.env | xargs) part, but still the same.

Anyone running django-q cluster with supervisor could share its configuration?
I'll share my mine here if I get something to work.

[GabLeRoux]

Ah I think I found it.

It seems to work fine with stopasgroup = true.

Final config looks like this:
/etc/supervisor/conf.d/django-q.conf

[program:django-q]
command = /srv/project-django-q-runner.sh
user = vagrant
stopasgroup = true

/srv/project-django-q-runner.sh

#!/usr/bin/env bash

export $(cat {{api.path}}/.env | xargs) 
/usr/local/virtualenvs/project/bin/python /srv/project/manage.py qcluster

From the docs:

If true, the flag causes supervisor to send the stop signal to the whole process group and implies killasgroup is true. This is useful for programs, such as Flask in debug mode, that do not propagate stop signals to their children, leaving them orphaned.

I'll send a PR to update documentation.