Subject: Re: Poor NFS error recovery
To: None <>
From: Greg Earle <earle@isolar.Tujunga.CA.US>
List: current-users
Date: 12/14/1994 11:39:45
>> I have an i386 machine with an old 8-bit 3c503 card; this card is
>> lame enough that it produces a handful of "ring buffer overruns"
>> under heavy network load.  This machine both serves and clients NFS.
>> Due (I presume) to the ethernet card, access (read/write) of file
>> larger than ~6K hangs.  (5.8K works, 6.2K fails).
> What are the exact sizes?  What about other sizes?
> Some else already suggested that you may be receiving more packets
> back-to-back than you have buffer space for, and suggested pulling back the
> rsize and wsize on the mount.
> I ran into a similar problem once, but it was much more infuriating.  Machine
> A would NFS mount a directory from machine B.  Trying to cat certain files -
> maddeningly few of them, but a given file either always worked or always
> failed - would hang, as if the server were unresponsive.

Not that this is related, but we just ran into these same symptoms recently,
and it turned out that it was happening because our Cisco Catalyst Ethernet
switches get really unhappy with Solaris 2.3 systems that were running with
MTU Discovery enabled.  The Solaris boxes were spitting out NFS packets onto
an FDDI ring with the Don't Fragment bit set, and of course the Catalyst
switches need to fragment them to forward them onto their switched Ethernet

I suspect this isn't the problem with the original poster, but there can be
more than one reason why small NFS traffic works and "regular" (i.e., 8k)
packets don't.

	- Greg