docker-celery

This is a Docker image for Python Celery workers.

Details

• Built on Alpine Linux for a super small image size.
• The Celery process is managed by Supervisor, and is pointed to the application through an environment variable.
• Other environment variables control the queue name, number of workers, and log level.
• Uses an external broker so the number of workers can be scaled up by simply scaling the number of containers.

Quick start example

This will show you how to get the example application running.

In order to get started, you will need a message broker running, such as RabbitMQ. If you don't already have one, you can get a free RabbitMQ instance for development purposes through CloudAMQP. It should only take a few minutes to set up.

Once you have one, take note of the AMQP URL, which will look something like this: amqp://****:*****@donkey.rmqp.cloudamqp.com/****.

Now, if you haven't already, clone this repository and pull the Docker image:

git clone https://github.com/epwalsh/docker-celery && cd docker-celery
docker pull epwalsh/docker-celery

Next, cd into the example/ directory and create the environment file access.txt with the AMQP URL:

cd example
cp access.sample.txt access.txt

Edit the first line in access.txt so that BROKER_URL is set to your AMQP URL. Now build and run the image with

docker build -t YOUR-NAME/celery-test .
docker run --env-file=./access.txt --rm YOUR-NAME/celery-test

If everything works, the Celery worker should start and produce some messages like this:

[2018-05-26 21:33:29,024: INFO/MainProcess] Connected to amqp://****:**@donkey.rmq.cloudamqp.com:****/****
[2018-05-26 21:33:29,582: INFO/MainProcess] mingle: searching for neighbors
[2018-05-26 21:33:31,737: INFO/MainProcess] mingle: all alone
[2018-05-26 21:33:32,877: INFO/MainProcess] test_queue-worker@4e39494c2213 ready.

You can now send a task to your worker with a Python script like this:

from celery import Celery

celery_app = Celery("example_app", broker="amqp://****")  # `broker` should be your BROKER_URL from access.txt
celery_app.send_task("worker.tasks.hello_world", queue="test_queue")