GitHub user javicacheiro opened a pull request:
https://github.com/apache/zookeeper/pull/413
WriteLock recipe: Fix bug in znode ordering when the sessionId is part of
the znode name
This PR solves a bug in the WriteLock recipe related to the ordering the
znodes.
In the original version when the nodes are sorted in WriteLock.java using a
TreeSet the whole znode path is taken into account and not just the sequence
number.
This causes an issue when the sessionId is included in the znode path
because a znode with a lower sessionId will appear as lower than other znode
with a higher sessionId even if its sequence number is bigger. In specific
situations this ended with two clients holding the lock at the same time.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/javicacheiro/zookeeper fix_writelock
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/zookeeper/pull/413.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #413
----
commit f9c581d4d92136f1ab718f5c24ae836c1c93dd03
Author: Javier Cacheiro <[email protected]>
Date: 2017-11-03T14:24:36Z
Fix bug in znode ordering when the sessionId is part of the znode name
----
---
