Message-ID: <503B49E4.2050409@tudelft.nl>
Date: Mon, 27 Aug 2012 12:20:20 +0200
From: Robbert Eggermont <R.Eggermont@tudelft.nl>
MIME-Version: 1.0
To: linux-nfs@vger.kernel.org
Subject: Re: nfsv4 state id errors
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Sender: linux-nfs-owner@vger.kernel.org

Hello,

> A couple of weeks agoo we upgraded our Netapp filer to 8.1.1RC1. We have
> always had trouble with NFS and state id's. Resulting in hanging clients.

(Client: SLED11 SP2, 3.0.34-0.7-default x86_64)

The "hanging" seems to be caused by the server returning an 
NFS4ERR_BAD_STATEID(10025) status for PUTFH;LOCK and PUTFH;WRITE;GETATTR 
calls. The client is making several hundreds of these identical (same 
filehandle) calls each second.

The problem seems to be triggered by (write) actions on an open file 
when the Kerberos ticket has expired (especially for .xsession-errors 
and files opened by Firefox or Thunderbird).

When (after some time) a new valid ticket is obtained, the kernel 
reports this error:
nfs4_reclaim_open_state: unhandled error -10026. Zeroing state

Is this a client or a server problem (and is there by any chance a fix)?

Robbert