#celery

I had trouble using rabbitmq cluster behind a loadbalancer. But not sure if you are having same issues as I did ...

The trick that worked for me was disabling connection pooling ...

this is shell output: http://dpaste.com/3QXRMQJ

it is able to connect to the cluster

Yeah, but doesnt it loose the connection at some point?

Djarnis: I have no idea :(

shouldn't be worker running always

Yes.

then why it would lose connection?

I think there is some bug in the connection pooling when rabbitmq is behind a load balancer … But you dont get ANY tasks through?

I have only one task :|

its also a single queue

Yes, but rabbitmq is in a cluster?

Djarnis: I am not sure. I have to check. If it helps, I am using the free instance in Lemur plan. https://www.cloudamqp.com/plans.html

i doubt if thats a cluster

Djarnis: does this code looks fine: http://dpaste.com/0T0K2SQ

I am doubting on `send_task`

I am using .delay when handing tasks to celery ...

Ahh … I guess it’s because it’s not registrered in the current process ...

av_: This is how my setup looks like ish https://gist.github.com/Djarnis/c0d0df156568101...

Djarnis: yup

av_: Just updated with settings.py as well ;)

I am using the default settings as show in docs: http://docs.celeryproject.org/en/latest/django/...

Try calling celery_app.config_from_object('django.conf:settings') after your celery_app = Celery('tasks', broker=settings.BROKER_URL) … ?

Djarnis: thats in tornado code

different code base

ok … Don’t know tornado ;/ So I guess someone else should take over from here ;)

I am sure its not issue from tornado. cos tornado code is able to put the message in queue (I can see admin panel of CloudAMPQ)

its with the django code which is not able to consume the message

av_: And you also put the celery code in your __init__.py?

yes

And you get nothing from the logs?

this is my __init__.py http://dpaste.com/3K842QR

Djarnis: logs of celery worker right? here are they: http://dpaste.com/3QXRMQJ

Yeah, it looks alright ...

yeah :(

Djarnis: I found out why. I am putting in a queue called `inbound` and the consumer task is not connected.

how do I specify this in Djanfo

just ommit the queue? It uses default celery queue? ;)

Yeah, that is probably the fix ;)

yeah

so how do I make the worker task to listen to that specific queue?

celery? It should fetch from the celery queue … If you don’t specify queue in your call, I am sure you’ll be fine ;)

I'm unable to run celery in beat mode, the command exits immediately: https://dpaste.de/Xw7j

Any idea how to debug this?

Djarnis: I meant, Iam not using the default queue. my queue name is `inbound`

so how do I configure my django code to consume this queue

i think `CELERY_QUEUES `

celery -A proj worker -l info -Q inbound?

If the queue name is defined in CELERY_QUEUES it will use that configuration

this is what I did: http://dpaste.com/248K87C

and it worked :D

av_: Awesome! ;)

Thanks a lot! :D

Glad I could “help” ;)

if this was not working, I had to deal with Pika. Man that one is pita to config

Djarnis: :D

av_: ;)