Dead simple task queue using redis. GitHub.

# tasks.py
import dsq
manager = dsq.create_manager()

def add(a, b):
    print a + b

if __name__ == '__main__':
    add.push(1, 2)
$ python tasks.py
$ dsq worker -bt tasks normal

See Tutorial for introduction.


The goal is a simple design. There is no worker manager, one can use supervisord/circus/whatever to spawn N workers. Simple storage model. Queue is a list and scheduled tasks are a sorted set. There are no task keys. Tasks are items of list and sorted set. There is no any registry to manage workers, basic requirements (die after some lifetime and do not hang) can be handled by workers themselves. Worker do not store result by default.