This document describes the current stable version of Celery (3.1). For development docs, go here.

Using SQLAlchemy

Experimental Status

The SQLAlchemy transport is unstable in many areas and there are several issues open. Unfortunately we don’t have the resources or funds required to improve the situation, so we’re looking for contributors and partners willing to help.



Celery needs to know the location of your database, which should be the usual SQLAlchemy connection string, but with ‘sqla+’ prepended to it:

BROKER_URL = 'sqla+sqlite:///celerydb.sqlite'

This transport uses only the BROKER_URL setting, which have to be an SQLAlchemy database URI.

Please see SQLAlchemy: Supported Databases for a table of supported databases.

Here’s a list of examples using a selection of other SQLAlchemy Connection Strings:

# sqlite (filename)
BROKER_URL = 'sqla+sqlite:///celerydb.sqlite'

# mysql
BROKER_URL = 'sqla+mysql://scott:tiger@localhost/foo'

# postgresql
BROKER_URL = 'sqla+postgresql://scott:tiger@localhost/mydatabase'

# oracle
BROKER_URL = 'sqla+oracle://scott:tiger@'


To store results in the database as well, you should configure the result backend. See Database backend settings.


The SQLAlchemy database transport does not currently support:

  • Remote control commands (celery events command, broadcast)
  • Events, including the Django Admin monitor.
  • Using more than a few workers (can lead to messages being executed multiple times).