txredlock - Redis distributed locks in Python - Now with more deferreds!

Warning: This project is incomplete. Although you can acquire a lock through a deferred, the Deferred will still contact each Redis server one by one. I would like to change this so that it attempts to acquire locks on all Redis servers simultaneously.

• Update CLI to use Twisted
• Make locks contact servers concurrently

This twisted python lib implements the Redis-based distributed lock manager algorithm described in this blog post.

To create a lock manager:

@inlineCallbacks
def setupRedlock():
	dlm = Redlock([{"host": "localhost", "port": 6379, "db": 0}, ])
	yield dlm.connect()

To acquire a lock:

@inlineCallbacks
def getLock():
	my_lock = yield dlm.lock("my_resource_name",1000)

Where the resource name is an unique identifier of what you are trying to lock and 1000 is the number of milliseconds for the validity time.

The returned value is False if the lock was not acquired (you may try again), otherwise an namedtuple representing the lock is returned, having three fields:

• validity, an integer representing the number of milliseconds the lock will be valid.
• resource, the name of the locked resource as specified by the user.
• key, a random value which is used to safe reclaim the lock.

To release a lock:

@inlineCallbacks
def releaseLock(my_lock):
	yield dlm.unlock(my_lock)

It is possible to setup the number of retries (by default 3) and the retry delay (by default 200 milliseconds) used to acquire the lock.

Both dlm.lock and dlm.unlock raise a exception MultipleRedlockException if there are errors when communicating with one or more redis masters. The caller of dlm should use a try-catch-finally block to handle this exception. A MultipleRedlockException object encapsulates multiple redis-py.exceptions.RedisError objects.

Disclaimer: This code implements an algorithm which is currently a proposal, it was not formally analyzed. Make sure to understand how it works before using it in your production environments.

Further Reading: http://redis.io/topics/distlock http://martin.kleppmann.com/2016/02/08/how-to-do-distributed-locking.html http://antirez.com/news/101 https://medium.com/@talentdeficit/redlock-unsafe-at-any-time-40ceac109dbb#.uj9ffh96x