From owner-freebsd-questions  Tue Dec 23 12:18:23 1997
Return-Path: <owner-freebsd-questions>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id MAA04418
          for questions-outgoing; Tue, 23 Dec 1997 12:18:23 -0800 (PST)
          (envelope-from owner-freebsd-questions)
Received: from conductor.synapse.net (conductor.synapse.net [199.84.54.18])
          by hub.freebsd.org (8.8.7/8.8.7) with SMTP id MAA04402
          for <freebsd-questions@freebsd.org>; Tue, 23 Dec 1997 12:18:17 -0800 (PST)
          (envelope-from evanc@synapse.net)
Received: (qmail 16677 invoked from network); 23 Dec 1997 20:17:44 -0000
Received: from piano.synapse.net (199.84.54.22)
  by conductor.synapse.net with SMTP; 23 Dec 1997 20:17:44 -0000
Date: Tue, 23 Dec 1997 15:17:44 -0500 (EST)
From: Evan Champion <evanc@synapse.net>
To: Tom <tom@uniserve.com>
cc: freebsd-questions@freebsd.org
Subject: Re: Improving NFS Performance
In-Reply-To: <Pine.BSF.3.96.971223120247.16399C-100000@shell.uniserve.com>
Message-ID: <Pine.BSF.3.96.971223151414.24258A-100000@piano.synapse.net>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-freebsd-questions@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

On Tue, 23 Dec 1997, Tom wrote:

>   For NFSv2, were you using udp or tcp mounts?

In both cases, I was using udp.  My options were:

rw,soft,intr,bg

> - Poor options for nfsv2
> - Bad interaction with network card.  NFS udp can overrun some cards too
> easily, meaning the card is always dropping packets.  I belive tcp mounts
> will back off better.

It made absolutely no difference whether or not I used tcp mounts.  I was
using identical options, with the exception that for NFS version 2 I
specified 'nfsv2'.  I tried using the small blocksize (rsize/wsize=1024),
and it didn't help nfsv3 at all.

So how is it that if both are running UDP with identical options, that
nfsv2 is about 100 times faster than nfsv3 on my net? :-)

Evan