This document describes Celery 2.2. For development docs, go here.

First steps with Celery

Creating a simple task

In this tutorial we are creating a simple task that adds two numbers. Tasks are defined in normal Python modules.

By convention we will call our module, and it looks like this:
from celery.task import task

def add(x, y):
    return x + y

Behind the scenes the @task decorator actually creates a class that inherits from Task. The best practice is to only create custom task classes when you want to change generic behavior, and use the decorator to define tasks.

See also

The full documentation on how to create tasks and task classes is in the Tasks part of the user guide.


Celery is configured by using a configuration module. By default this module is called

The configuration module must either be in the current directory or on the Python path, so that it can be imported.

You can also set a custom name for the configuration module by using the CELERY_CONFIG_MODULE environment variable.

Let’s create our

  1. Configure how we communicate with the broker (RabbitMQ in this example):

    BROKER_HOST = "localhost"
    BROKER_PORT = 5672
    BROKER_USER = "myuser"
    BROKER_PASSWORD = "mypassword"
    BROKER_VHOST = "myvhost"
  2. Define the backend used to store task metadata and return values:


    The AMQP backend is non-persistent by default, and you can only fetch the result of a task once (as it’s sent as a message).

    For list of backends available and related options see Task result backend settings.

  3. Finally we list the modules the worker should import. This includes the modules containing your tasks.

    We only have a single task module,, which we added earlier:

    CELERY_IMPORTS = ("tasks", )

That’s it.

There are more options available, like how many processes you want to use to process work in parallel (the CELERY_CONCURRENCY setting), and we could use a persistent result store backend, but for now, this should do. For all of the options available, see Configuration and defaults.


You can also specify modules to import using the -I option to celeryd:

$ celeryd -l info -I tasks,handlers

This can be a single, or a comma separated list of task modules to import when celeryd starts.

Running the celery worker server

To test we will run the worker server in the foreground, so we can see what’s going on in the terminal:

$ celeryd --loglevel=INFO

In production you will probably want to run the worker in the background as a daemon. To do this you need to use the tools provided by your platform, or something like supervisord (see Running celeryd as a daemon for more information).

For a complete listing of the command line options available, do:

$  celeryd --help

Executing the task

Whenever we want to execute our task, we use the delay() method of the task class.

This is a handy shortcut to the apply_async() method which gives greater control of the task execution (see Executing Tasks).

>>> from tasks import add
>>> add.delay(4, 4)
<AsyncResult: 889143a6-39a2-4e52-837b-d80d33efb22d>

At this point, the task has been sent to the message broker. The message broker will hold on to the task until a worker server has consumed and executed it.

Right now we have to check the worker log files to know what happened with the task. This is because we didn’t keep the AsyncResult object returned.

The AsyncResult lets us check the state of the task, wait for the task to finish, get its return value or exception/traceback if the task failed, and more.

Let’s execute the task again – but this time we’ll keep track of the task by holding on to the AsyncResult:

>>> result = add.delay(4, 4)

>>> result.ready() # returns True if the task has finished processing.

>>> result.result # task is not ready, so no return value yet.

>>> result.get()   # Waits until the task is done and returns the retval.

>>> result.result # direct access to result, doesn't re-raise errors.

>>> result.successful() # returns True if the task didn't end in failure.

If the task raises an exception, the return value of result.successful() will be False, and result.result will contain the exception instance raised by the task.

Where to go from here

After this you should read the User Guide. Specifically Tasks and Executing Tasks.

Previous topic

Broker Installation

Next topic


This Page