From owner-freebsd-questions  Fri Aug 28 17:54:36 1998
Return-Path: <owner-freebsd-questions@FreeBSD.ORG>
Received: (from majordom@localhost)
          by hub.freebsd.org (8.8.8/8.8.8) id RAA18043
          for freebsd-questions-outgoing; Fri, 28 Aug 1998 17:54:36 -0700 (PDT)
          (envelope-from owner-freebsd-questions@FreeBSD.ORG)
Received: from lucy.bedford.net (lucy.bedford.net [206.99.145.54])
          by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id RAA18036
          for <questions@FreeBSD.ORG>; Fri, 28 Aug 1998 17:54:32 -0700 (PDT)
          (envelope-from listread@lucy.bedford.net)
Received: (from listread@localhost)
	by lucy.bedford.net (8.8.8/8.8.8) id UAA20159;
	Fri, 28 Aug 1998 20:37:57 -0400 (EDT)
	(envelope-from listread)
Message-Id: <199808290037.UAA20159@lucy.bedford.net>
Subject: Re: NFS3 serving: FreeBSD vs. Irix
In-Reply-To: <199808251746.NAA16861@xxx.video-collage.com> from Mikhail Teterin at "Aug 25, 98 01:46:34 pm"
To: mi@video-collage.com (Mikhail Teterin)
Date: Fri, 28 Aug 1998 20:37:57 -0400 (EDT)
Cc: questions@FreeBSD.ORG
X-no-archive: yes
Reply-to: djv@bedford.net
From: CyberPeasant <djv@bedford.net>
X-Mailer: ELM [version 2.4ME+ PL38 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-questions@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

Mikhail Teterin wrote:
> Hello!
> 
> I'm seeing my PPro200 with FreeBSD 2.2.7-STABLE being about 3 times slower
> (according to iozone) then Irix 5.3 running on R4000 at 100MHz with SGI's
> latest NFS3 patches. Memory is plentyfull.
> 
> The client is also an Irix machine very similar to the server. The network
> is 10BaseT and the numbers are (24Mb test file):
> 
> 	Irix to FreeBSD:
> 
>         237929 bytes/second for writing the file
> 	5642561 bytes/second for reading the file
> 
> 	Irix to Irix:
> 
> 	717997 bytes/second for writing the file
> 	5423668 bytes/second for reading the file
> 
 ...snip...

DId you get any reples to this?

All I can think of is that FreeBSD and Irix are disagreeing
about the optimal block size. You might try forcing the r and
w blocksizes on the client to various values. (1024,2048,...).

Serving from PPro 200 to 486/33 8MB over 10Base2, I got consistent
write times of around 500000 bytes/second. (Both boxes running
FreeBSD 2.2.6-R).

Could there be an evil interaction of NFS block size and ethernet
packet size? (Like in Linux NFS?)

As a last resort you might try running tcpdump and seeing what's
going on. Does nfsstat tell you anything?

D.V.

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message