Question

I am writing a small Django application and I should be able to create
for each model object its periodical task which will be executed with
a certain interval. I’m use for this a Celery application, but i can’t understand one thing:

class ProcessQueryTask(PeriodicTask):
   run_every = timedelta(minutes=1)

   def run(self, query_task_pk, **kwargs):
       logging.info('Process celery task for QueryTask %d' %
query_task_pk)
       task = QueryTask.objects.get(pk=query_task_pk)
       task.exec_task()
       return True

Then i’m do following:

>>> from tasks.tasks import ProcessQueryTask
>>> result1 = ProcessQueryTask.delay(query_task_pk=1)
>>> result2 = ProcessQueryTask.delay(query_task_pk=2)

First call is success, but other periodical calls returning the error
– TypeError: run() takes exactly 2 non-keyword arguments (1 given) in
celeryd server.
Can I pass own params to PeriodicTask run()?

Answer 1

This was answered wonderfully by Ask Solem in his response to your question on the celery-users Google group.

Periodic tasks doesn’t use arguments, so you need to make several
classes or make a periodic task that processes more than one “model”.

E.g.:

from celery.task import PeriodicTask
from celery.decorators import periodic_task

# base class
class BaseProcessQueryTask(PeriodicTask):
    abstract = True
    run_every = timedelta(minutes=1)
    query_task_pk  = None

    def run(self):
        task = QueryTask.objects.get(pk=self.query_task_pk)
        task.exec_task()

class ProcessQueryTask1(BaseProcessQueryTask):
    query_task_pk = 1

class ProcessQueryTask2(BaseProcessQueryTask):
    query_task_pk = 2

but it’s more likely you want something like this:

@task(ignore_result=True)
def execute_query_task(task):
    task.exec_task()

@periodic_task(run_every=timedelta(minutes=1))
def process_query_tasks():
    for task in QueryTask.objects.all():
        ExecuteQueryTask.delay(task)