xfs
[Top] [All Lists]

Re: [RESOLVED] Re: XFS internal error

To: Nicolas Kowalski <Nicolas.Kowalski@xxxxxxx>
Subject: Re: [RESOLVED] Re: XFS internal error
From: James Pearson <james-p@xxxxxxxxxxxxxxxxxx>
Date: Fri, 16 May 2003 16:49:13 +0100
Cc: linux-xfs@xxxxxxxxxxx
Organization: Moving Picture Company
References: <vqowugrb4o2.fsf@xxxxxxxxxxxxxxx> <vqok7crf8d6.fsf@xxxxxxxxxxxxxxx> <vqoptmj9cf7.fsf_-_@xxxxxxxxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
I don't know if/how this is related to XFS, but if you change the SCSI
ID of a disk during a reboot that has NFS client mounts, then you will
get a Stale NFS file handle on the client after a server reboot. A way
round this is to use the relatively new fsid=N export option in
/etc/exportfs - in this way you can change the SCSI ID without the
clients caring. You need to be running the latest nfs-utils (1.0.3) and
have a recent kernel (2.4.21-rc2 should have support) to use this
option.

James Pearson

Nicolas Kowalski wrote:
> 
> Nicolas Kowalski <Nicolas.Kowalski@xxxxxxx> writes:
> 
> > Nicolas Kowalski <Nicolas.Kowalski@xxxxxxx> writes:
> >
> >> Hello.
> >>
> >> I am experiencing a very weird behaviour on our main fileserver,
> >> running a 2.4.21-rc2-xfs (cvs yesterday) :
> 
> I finally found what was wrong. One of our ~30 Linux NFS clients, with
> a "Stale NFS file handle" which occured during the server reboot was
> causing the errors on the Server.
> 
> I still think this is a weird behaviour (nfsd or xfs related, I do not
> know), but this does not matter, as I know now how to resolve it.
> 
> Sorry for the noise.
> 
> --
> Nicolas


<Prev in Thread] Current Thread [Next in Thread>