Part 5 - Using Celery to Automate Maintenance Chores¶
Scenario¶
In Part 1 you created your own Authorization Server and it’s running along just fine. However, the database is getting cluttered with expired tokens. You can periodically run the cleartokens management command, but why not automate this with Celery?
Set up RabbitMQ¶
Celery components communicate via a message queue. We’ll use RabbitMQ.
Install RabbitMQ on MacOS¶
If you are using MacOS it’s likely you are already using Homebrew. If not, now’s the time to install this fantastic package manager.
brew install rabbitmq
brew service start rabbitmq
Install RabbitMQ with Docker¶
This will start up a docker image that just works:
docker run -it --rm --name rabbitmq -p 5672:5672 -p 15672:15672 rabbitmq:3.9-management
Install RabbitMQ on Windows¶
See the RabbitMQ Installing on Windows instructions.
Add Celery¶
Make sure you virtualenv is active and install celery and django-celery-beat.
pip install celery django-celery-beat
Update your list of installed apps to include both your Authorization Server app – we’ll call it tutorial
,
and django_celery_beat
which extends your Django project to store your periodic task schedule
in the database and adds a Django Admin interface for configuring them.
INSTALLED_APPS = {
# ...
"tutorial",
"django_celery_beat",
}
Now add a new file to your app to add Celery: tutorial/celery.py
:
import os
from celery import Celery
# Set the default Django settings module for the 'celery' program.
os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'tutorial.settings')
app = Celery('tutorial', broker="pyamqp://guest@localhost//")
app.config_from_object('django.conf:settings', namespace='CELERY')
# Load task modules from all registered Django apps.
app.autodiscover_tasks()
This will autodiscover any tasks.py
files in the list of installed apps.
We’ll add ours now in tutorial/tasks.py
:
from celery import shared_task
@shared_task
def clear_tokens():
from oauth2_provider.models import clear_expired
clear_expired()
Finally, update tutorial/__init__.py
to make sure Celery gets loaded when the app starts up:
from .celery import app as celery_app
__all__ = ('celery_app',)
Run Celery Beat and the Worker¶
RabbitMQ should already be running; it’s the “glue” between Beat and the Worker.
It’s best to run each of these in its own terminal window so you can see the log messages.
Start Celery Beat¶
celery -A tutorial beat -l INFO --scheduler django_celery_beat.schedulers:DatabaseScheduler
Start Celery Worker¶
celery -A tutorial worker -l INFO
Configure the clear_tokens
task¶
Go into Django Admin and you’ll see a new section for periodic tasks:

Now let’s define a fairly short (10 second) interval. Go to: http://127.0.0.1:8000/admin/django_celery_beat/intervalschedule/ and select Add Interval, set number of intervals to 10 and interval period to seconds and Save.
Then go to http://127.0.0.1:8000/admin/django_celery_beat/periodictask/ to add a new periodic task by
selecting Add Periodic Task and
select tutorial.tasks.clear_tokens
, choose the every 10 seconds
interval schedule, and “Save.”

Now your Celery Beat and Celery Workers should start running the task every 10 seconds.
The Beat console will look like this:
[2022-03-19 22:06:35,605: INFO/MainProcess] Scheduler: Sending due task clear stale tokens (tutorial.tasks.clear_tokens)
And the Workers console like this:
[2022-03-19 22:06:35,614: INFO/MainProcess] Task tutorial.tasks.clear_tokens[5ec25fb8-5ce3-4d15-b9ad-750b80fc07e0] received
[2022-03-19 22:06:35,616: INFO/ForkPoolWorker-8] refresh_expire_at is None. No refresh tokens deleted.
[2022-03-19 22:06:35,629: INFO/ForkPoolWorker-8] 0 Expired access tokens deleted
[2022-03-19 22:06:35,631: INFO/ForkPoolWorker-8] 0 Expired grant tokens deleted
[2022-03-19 22:06:35,632: INFO/ForkPoolWorker-8] Task tutorial.tasks.clear_tokens[5ec25fb8-5ce3-4d15-b9ad-750b80fc07e0] succeeded in 0.016124433999999965s: None
References¶
The preceding is based on these references:
https://docs.celeryq.dev/en/stable/django/first-steps-with-django.html
https://docs.celeryq.dev/en/stable/userguide/periodic-tasks.html#beat-custom-schedulers
https://django-celery-beat.readthedocs.io/en/latest/index.html