LockerRoom – Distributed lock manager in Python

A problem when running systems in a network environment is that you might need to control access to resources so that only one of your programs updates a resource at a time.

For this you could use a Distributed Lock Manager (DLM) which provides synchronised access to resources over a network. If you are running Memcache or Redis you can find implementations for DLM:s here and here. But what if you want to use MongoDB for the task? Then LockerRoom might be for you!

LookerRoom is a simple implementation of a DLM that saves the locks in MongoDB, in that way making them easily accessible on a network.

You initialise the lock manager like this:

import locker_room
locker = locker_room.LockerRoom(host='server1')

Where host is hostname of the server running MongoDB. You can also specify which database and which collection to store the locks in if you don’t want the default values.

To get a lock you are recommended to either use locker as a context or a function decorator:

with locker.lock_and_release('my_lock', owner='gustav', timeout=2):
    # do important stuff


def important_function():
    # do important stuff

But it is also possible to just call lock and release separately:

locker.lock('my_lock', timeout=2)
# do stuff

The lock methods take optional values for timeout (how long to wait for accessing the lock before raising exception) and for owner (who owns the lock).

You can also see the status of a lock, i.e if it is locked, when the lock was set and who owns it, by calling the status method:

>> {u'owner': u'gustav', u'timestamp': datetime.datetime(2014, 4, 17, 14, 6, 8, 291000), 
    u'_id': u'my_lock',  u'locked': True}

The complete source code can be found here.