Comment 25 for bug 1557593

Reviewed: https://review.openstack.org/478131
Committed: https://git.openstack.org/cgit/openstack/tooz/commit/?id=1b5b21e7c01998bbb972920ec869e51d1c20f105
Submitter: Jenkins
Branch: stable/newton

commit 1b5b21e7c01998bbb972920ec869e51d1c20f105
Author: Julien Danjou <email address hidden>
Date: Sat Mar 18 10:33:57 2017 +0100

    redis: fix concurrent access on acquire()

    Redis-py is not thread safe and in lock management, which means the following
    can also happen:
    1. lock.acquire()
    2a. lock.release() ---\_ at the same time for all 2. operations
    2b. lock.acquire() ---/
    2c. lock.heartbeat() /

    which means the lock can be when a lock is already acquired and in the process
    of being released and re-acquired in paralllel:
    1. acquire() acquire the lock (self._lock.acquire())
    2. release() release the lock (self._lock.release()) and remove it from
       self._lock.acquire() where it's not (discard() does not raise on non-existing
       element)
    3. acquire() put the lock in _acquired_locks
    4. heartbeat() iterates on _acquired_locks and try to extend a lock that indeed
       was acquired but was in the mean time released by Redis.

    Change-Id: Ib9549b44485cb15de312ec9dfa57b148cb45e2c9
    Closes-Bug: #1557593
    (cherry picked from commit 059f2ae5406a325017ad181ace1e5b7c5bff0d22)