I have a LAN of 8 SGIs (Indy, O2, Octane, Onyx RE2) and I'm trying to
set up a disk server with an intel box running RedHat Linux 6.0 with
kernel 2.2.5-22. The intel box also have a second network interface that
leads to a mainly PC LAN with very small traffic.
When the SGIs don't do large and constant access to the server,
everything works fine. But at some point (which I haven't discovered
yet) all network traffic stops. This happens usually under heavy load,
when several users simultaneously request file from the server, but
sometimes for no apparent reasons.
The only messages on the SGIs are
NFS2 read failed for server dupont: Timed out
NFS2 write failed for server dupont: Timed out
The only peculiar message given by dmesg on the linux box is
eth0: Transmit error, Tx status register 82.
I tried using NFS3 instead, using async NFS, I checked all cables,
plugs, other machines, patches, etc., but I couldn't find anything that
made it work correctly.
For all I know, it could be a problem with the SCSI cards inside the PC
but the only symptoms are related to network.
If anybody as an idea, or if someone has a configuration similar, i
would be happy to hear any suggestion.
Silicon Worlds S.A.
12, rue de Chatillon 75014 Paris France
Tel: +33 (01) 18.104.22.168
Fax: +33 (01) 22.214.171.124