Description of problem:
Most gluster commands takes a cluster wide lock, before proceeding to staging process. And after the commit stage, the cluster is unlocked. But if in between the node, from which the command was executed goes down, the lock is not released, and the whole cluster stays in a lockdown stage.


Version-Release number of selected component (if applicable):


How reproducible:
Everytime


Steps to Reproduce:
1. Attach gdb to a glusterd process on one of the nodes, and put a breakpoint.
2. Run a gluster volume command from any other node.
3. While the process is waiting on the breakpoint, kill the node from which the command was given.

Actual results:
The whole cluster remains in a lockdown


Expected results:
The cluster should recover itself from the locked stage


Additional info: