restart doesn't restart processes

[luislavena]

Hello,

I've been testing mongroup locally on my OSX 10.8 box before deploying this to production and encountered a minor issue with restart.

I have this basic config:

pids = tmp/pids
logs = log
worker = rake resque:work QUEUE=main,ts_delta
apn = rake apn:sender QUEUE=apple_push_notifications

When started with mongroup, everything is ok:

$ mongroup -c config/mongroup.conf status

      worker : 4093 : alive : uptime 7 seconds
         apn : 4096 : alive : uptime 7 seconds

But when done a restart

$ mongroup -c config/mongroup.conf restart

        stop : worker
        stop : apn
     running : worker
     running : apn

The child processes are terminated and never get up again, so status reports them dead:

$ mongroup -c config/mongroup.conf status

      worker : 4093 : dead
         apn : 4096 : dead

And using ps to verify I confirm they are dead 👅

Last time I checked resque did work properly for this (used with foreman and Procfiles) so I think it respond properly to process termination signals.

Thank you for creating this tool! ❤️ ❤️ ❤️

[luislavena]

Forgot to mention that as a workaround I'm doing:

mongroup -c config/mongroup.conf stop && sleep 5 && mongroup -c config/mongroup.conf start

Resque takes time to terminate, so perhaps that is causing the issue with restart?

[jgallen23]

I think it's because of the ='s in your config file. There is a pull request open that should fix that

[matthewmueller]

+1, running into this as well. Also have an = in the config file for passing vars through.

[jgallen23]

try now. I just published 0.4.0

[juliangruber]

I have this problem too and I'm pretty sure it's because the process takes too long to quit.

what about adding wait $PID after the kill $PID in stop?

[juliangruber]

when mon is killed with SIGQUIT it dies before its watched process dies...

However, if i change the sleep in restart to wait for 5 seconds it works. So, the time the process takes to exit is the cause for the restart not working.