[DRBD-user] Stale NFS file handle problem

[Mike]

Hello list,

	I am working with an idential pair of boxes running debian squeeze and 
have followed howto's to configure them up as master/slave replicated 
with drbd + heartbeat. For the most part, everything works, replication 
is happening, I can shut down the primary and the file system is still 
available and I can go back and forth. However, after a switch, I get 
'stale nfs file handle' errors for any files being written to at the 
time the switch happens but this appears to be only for those files any 
any new files touched/created have no such issue.

	I have 'fsid=1' in both /etc/exports. I am mounting the nfs server with 
UDP. I have /var/lib/nfs on the shared file system. I have -t servername 
on my rpc options. I have rechecked all of these on both servers. The 
file system in question is an LVM volume (/dev/vg00/replicated) 
containing a btrfs file system. I have looked at every google result 
about this issue and nothing has seemed to help. I've changed the 
nfs-kernel-server script to kill with -9 and I removed the 'exportfs 
-au' lines as well. No dice.

	Any help out there?

	Thanks.

Mike-