Simple job queues for Python
Go to file
Vincent Driessen a491ff248d Add simple helper function, to avoid users to learn about the `conn`
stack.

This makes it easier to start using the product.
2011-11-17 13:29:32 +01:00
examples Update README (and example). 2011-11-15 22:45:51 +01:00
rq Add simple helper function, to avoid users to learn about the `conn` 2011-11-17 13:29:32 +01:00
tests Add equality and comparison methods. 2011-11-16 12:45:16 +01:00
.gitignore Add some project meta stuff. 2011-11-14 12:11:58 +01:00
README.md Update README with some project values. 2011-11-16 00:53:56 +01:00
run_tests Add some project meta stuff. 2011-11-14 12:11:58 +01:00
setup.py Make it an actual PyPI-managable Python package. 2011-11-14 12:16:28 +01:00

README.md

WARNING: DON'T USE THIS IN PRODUCTION (yet)

RQ: Simple job queues for Python

RQ is a lightweight Python library for queueing work and processing them in workers. It is backed by Redis.

Putting jobs on queues

To put jobs on queues, first declare a Python function to be called on a background process:

def slow_fib(n):
    if n <= 1:
        return 1
    else:
        return slow_fib(n-1) + slow_fib(n-2)

Notice anything? There's nothing special about a job! Any Python function can be put on an RQ queue, as long as the function is in a module that is importable from the worker process.

To calculate the 36th Fibonacci number in the background, simply do this:

from rq import Queue
from fib import slow_fib

# Calculate the 36th Fibonacci number in the background
q = Queue()
q.enqueue(slow_fib, 36)

If you want to put the work on a specific queue, simply specify its name:

q = Queue('math')
q.enqueue(slow_fib, 36)

You can use any queue name, so you can quite flexibly distribute work to your own desire. Common patterns are to name your queues after priorities (e.g. high, medium, low).

The worker

NOTE: You currently need to create the worker yourself, which is extremely easy, but RQ will include a custom script soon that can be used to start arbitrary workers without writing any code.

Creating a worker daemon is also extremely easy. Create a file worker.py with the following content:

from rq import Queue, Worker

q = Queue()
Worker(q).work_forever()

After that, start a worker instance:

python worker.py

This will wait for work on the default queue and start processing it as soon as messages arrive.

You can even watch several queues at the same time and start processing from them:

from rq import Queue, Worker

queues = map(Queue, ['high', 'normal', 'low'])
Worker(queues).work()

Which will keep popping jobs from the given queues, giving precedence to the high queue, then normal, etc. It will return when there are no more jobs left (contrast this to the previous example using Worker.work_forever(), which will never return since it keeps waiting for new work to arrive).

Installation

Simply use the following command to install the latest released version:

pip install rq

If you want the cutting edge version (that may well be broken), use this:

pip install -e git+git@github.com:nvie/rq.git@master#egg=rq

Project History

This project has been inspired by the good parts of Celery, Resque and this snippet, and has been created as a lightweight alternative to the heaviness of Celery or other AMQP-based queueing implementations.

Project values:

  • Simplicity over completeness
  • Fail-safety over performance
  • Runtime insight over static configuration upfront

This means that, to use RQ, you don't have to set up any queues up front, and you don't have to specify any channels, exchanges, or whatnot. You can put jobs onto any queue you want, at runtime. As soon as you enqueue a job, it is created on the fly.