Peter Hoffmann Director Data Engineering at Blue Yonder. Python Developer, Conference Speaker, Mountaineer

Simple Python Queue with Redis

The following article shows how to use redis to build a simple multi-producer, multi-consumer Queue with an interface similar to the python standardlib queue. With this queue you can easily share data between multiple processes or offload time consumig calculations to multiple worker processes.

To store the data we use the redis list data type. Redis Lists stores simple strings sorted by insertion order.

The following redis commands are used:

  • rpush Insert an element at the tail of the list
  • blpop Get an element from the head of the list, block if list is empty
  • lpop Get an element from the head of the list, return nothing list is empty
  • llen Return the length of the list

The implementation uses the redis-py library to talk to the server.

import redis

class RedisQueue(object):
    """Simple Queue with Redis Backend"""
    def __init__(self, name, namespace='queue', **redis_kwargs):
        """The default connection parameters are: host='localhost', port=6379, db=0"""
       self.__db= redis.Redis(**redis_kwargs)
       self.key = '%s:%s' %(namespace, name)

    def qsize(self):
        """Return the approximate size of the queue."""
        return self.__db.llen(self.key)

    def empty(self):
        """Return True if the queue is empty, False otherwise."""
        return self.qsize() == 0

    def put(self, item):
        """Put item into the queue."""
        self.__db.rpush(self.key, item)

    def get(self, block=True, timeout=None):
        """Remove and return an item from the queue. 

        If optional args block is true and timeout is None (the default), block
        if necessary until an item is available."""
        if block:
            item = self.__db.blpop(self.key, timeout=timeout)
            item = self.__db.lpop(self.key)

        if item:
            item = item[1]
        return item

    def get_nowait(self):
        """Equivalent to get(False)."""
        return self.get(False)


>>> from RedisQueue import RedisQueue
>>> q = RedisQueue('test')
>>> q.put('hello world')

Now if we have a look at the redis database with the redis-cli client it shows the expected results:

redis> keys *
1) "queue:test"
redis> type queue:test
redis> llen queue:test
(integer) 1
redis> lrange queue:test 0 1
1) "hello world"

We can get the item from a different script with:

>>> from RedisQueue import RedisQueue
>>> q = RedisQueue('test')
>>> q.get()
'hello world'

A subsequent call of q.get() will block until anotherone puts a new item into the Queue.

The next step would be to an endoder/decoder (e.g python-json) to the Queue so that you are not limited to send strings.

There alredy exists the nice and simple hotqueue library which has the same interface as the above example and provides encoding/decoding.

Other mentionable queue implementations with a redis backend are:

  • flask-redis A basic Message Queue with Redis for flask.
  • celery An asynchronous task queue/job queue based on distributed message passing. Much more advanced. Can be used with different storage backends.
  • rq Simple python library for queueing jobs and processing them in the background with workers.
  • resque is a Redis-backed Ruby library for creating background jobs, placing them on multiple queues, and processing them later. Used at github. Includes a nice monitoring web interface.
  • pyres A resque clone in python.