dlm: fix waiter recovery
authorDavid Teigland <teigland@redhat.com>
Mon, 23 Apr 2012 17:18:18 +0000 (12:18 -0500)
committerDavid Teigland <teigland@redhat.com>
Thu, 26 Apr 2012 20:36:04 +0000 (15:36 -0500)
An outstanding remote operation (an lkb on the "waiter"
list) could sometimes miss being resent during recovery.
The decision was based on the lkb_nodeid field, which
could have changed during an earlier aborted recovery,
so it no longer represents the actual remote destination.
The lkb_wait_nodeid is always the actual remote node,
so it is the best value to use.

Signed-off-by: David Teigland <teigland@redhat.com>

No differences found