#celery

hi

what could be the reason if my tasks doesn't detach

they arent asyncronious

i start workers

i add .delau(<params>)

i tryed with apply_async

can a celery task call another celery task?

hiya, I seem to have an issue where I want to stress test a worker by isolating it from the celery queue, but it seems to always listen to it

HexKey: short answer yes

HexKey: but there are constructs if you want them to be done in a certain order

i think order wont matter about order

HexKey: you can try http://docs.celeryproject.org/en/latest/usergui...

I think the link keyword

in a schedueled task i add some files to a database, the i want to trigger a lengthy analysis of those files in another task, because sometimes this task will be called in a scheduled manner also

HexKey: you have to model you processing so you don't wait for results of tasks (eg: .get() ) in other tasks

but if you really need to you would need a different worker pool + queue for each task

that really complicates stuff - don't do it like that

hmmm

the processes are totally disconnected in time, as long as the stuff is in the databse is does matter when the called task within the database persist task is executed

i dont think there is a problem with having to wait for things

ofcourse not

having a deadlock is a problem :)

as long as celery allows i think i should be safe just calling the sub-task after i persist into the databse

anyway, in regards to my workers listening to the celery queue no matter what, I did this:

CELERY_DEFAULT_QUEUE = 'host_telapi5_prd_nj1_teltechsys_net'

del CELERY_ROUTES

should that not wor?

also passed -Q custome_qeuue to the workers

(celery multi)

any ideas why it would still listen to the celery queue?

HexKey: maybe you're also interested in https://github.com/celery/celery/pull/817?

aimatt: what broker?

ionelmc: rabbit

aimatt: perhaps your arguments to celeryd are wrong

check that the command lines for celeryd are proper

run `celeryd_multi show <the old arguments>` to see the command lines

manage.py celeryd_multi start 16 -P gevent --autoscale=1,5 -Q custom_queue

hi guys

i'm trying to make this task work:

@periodic_task(run_every=crontab(hour=20, minute=14, day_of_week="mon"))

but it's not starting

b_52globe: make sure you start celerybeat

it's the daemon that checks the clock and puts the task into the broker

hm i start celery with : python manage.py celeryd -E -l info -c 2 -S djcelery.schedulers.DatabaseSchedulter --loglevel=info

aimatt, i must start another deamon ?

b_52globe: http://docs.celeryproject.org/en/latest/usergui...

aimatt, ok thks

I think there may be a flag to pass into celeryd if you only have one instance

b_52globe: the reason it is separate is because if you have multiple beats running they will get scheduled multipel times

celery worker -B

will start celeryd with beat embedded

aimatt, thanks

np

asksol you here

this is infuriating

still subscribes to the celery queue

guys how to define a periodic task but that accept parameter

from a django view for ex, call and pass to it parameters

b_52globe: not sure you can

nvm you can

just define it like a normal task

then in admin under admin/djcelery/periodictask/xx/

theres an arguments accordion

json encoded

aimatt, even i want the task to be launched in one week later

ok checking admin

b_52globe: what's the issue? define the task in tasks.py and go to admin to add it

aimatt, yeap thanks

np

b_52globe: a tip: when you import a task in another module, I recommend using the full import path as it determines the task name

aimatt, hm ok thanks

celeryd is disregarding -Q

celeryd_detach and celeryd_multi

asksol i want to add queue exclusions to celery workers, if you have thoughts on it let me know and ill send up a pull req (something like -X or something, similar to -Q)

anyone have any idea how to prevent celery from pulling tasks out of the default "celery" queue?

'CELERY_ALWAYS_EAGER': True,

i know the celery monitor in the django is deprecated but it's convenient for basic monitoring, but the admin shows my worker as offline even though the last heartbeat shown in the model detail view is a few seconds old

inside the django admin*

MVXA: try flower, celery monitor