state recovery failed on NFSv4 server with error 2

[Trond Myklebust]

On Thu, 2008-03-27 at 11:32 -0400, Thomas Garner wrote:
> Since it'll be a while before the error is triggered again, this is
> the first 15M of my first 2.6G dump during the error, created using
> the following tcpdump options on the client side:
> 
> tcpdump -s 1500 -w ~/nfs_dump_argento -x -i eth1
> 
> http://s120158928.onlinehome.us/nfs_dump_argento.001.bz2
> 
> Thomas

It looks as if the client is spewing new READ and WRITE requests without
handling the NFS4ERR_EXPIRED error returned by the server. AFAICS, that
can only happen if the recovery thread Oopses.

Does 'ps -efww' show a thread with a name that starts with the
IP-address of the server?
Do the syslogs, or does 'dmesg -s 90000' show an Oops involving that
process?

Cheers
  Trond