#celery

We are having an issue in one of our Celery systems in production. We've got 2 nodes, 4 processes in each one, consuming from a queue. Messages within the queue execute tasks that rely on an external system, they usually take a couple seconds but sometimes this system degrades and hard time limit of 300 seconds kicks in. After several hard time limits, processes of both nodes

stop consuming the queue and inspect active shows that they are not doing anything. Celery version is 3.1.11 and django-celery is 3.1.10. Against RabbitMQ 3.1.2 and py-amqp 1.4.5

I've been reviewing Celery bug tracking and apparently there is no open issue with this kind of environment, shall i report it? anything I can add that would be helfpul?

asksol_: is it possible to programatically run celeryd?

without using popen?

graingert_: http://docs.celeryproject.org/en/master/interna...

from celery.bin.worker import Worker?

that's if you want to use argv (or detach)

from my_package.app import app; app.worker_main();

graingert_, don't try to run it in ipython

malinoff: ?

graingert_, you will have to kill ipython process when celery worker will be stopped (by C-c)

asksol_: Have you seen what i've posted before ?

maraujop: what did you post?