On Sun, Jan 21, 2001 at 03:37:41PM -0600, Ruibiao Qiu wrote:
> On Sun, 21 Jan 2001, Manuel Bouyer wrote:
> 
> >On Sat, Jan 20, 2001 at 12:03:40PM -0600, Ruibiao Qiu wrote:
> >> Here are the tcpdump output ('tcpdump -vvv port 2049'):
> >
> >Well, could you retry with  'tcpdump -s 1500 -vvv port 2049' ?
> >I think we are missing some infos because the data collected were
> >truncated.
> 
> It is attached below.
> 
> >Still, I have several 1.5 and later client talking with 1.4.x and 1.3.x servers
> >without troubles.
> 
> I know it works cuz we have been always using it, but there was
> some recent changes to the system (a cutomerized development
> kernel with binaries, but I dont think it disturbs the NFS)  We
> just wnat to figure out what we changed incorrectly.  
> 
> tcpdump: listening on ex0
> 15:05:08.711541 client.2278043160 > server.nfs: 92 fsinfo [|nfs] (ttl 64, id 11167)
> 15:05:08.711776 server.nfs > client.2278043160: reply ok 32 fsinfo ERROR: 'Stale NFS file handle' POST: (ttl 64, id 22508)
> 15:05:08.712257 client.2278043161 > server.nfs: 92 fsstat fh 4,10/1931 (ttl 64, id 11168)
> 15:05:08.712359 server.nfs > client.2278043161: reply ok 32 fsstat ERROR: 'Input/output error' POST: [|nfs] (ttl 64, id 22509)

Ok, so the I/O error comes from the server. Now need to know why the server
returns such error ... Sorry, but I don't have much idea.
Do the server have multiple interfaces ?

--
Manuel Bouyer <bouyer@antioche.eu.org>
--