From owner-freebsd-net Sun Dec 22 20: 5:30 2002 Delivered-To: freebsd-net@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id BEB0937B406 for ; Sun, 22 Dec 2002 20:05:28 -0800 (PST) Received: from goof.com (pcp02305702pcs.longhl01.md.comcast.net [68.52.164.8]) by mx1.FreeBSD.org (Postfix) with SMTP id A54F843EEC for ; Sun, 22 Dec 2002 20:05:27 -0800 (PST) (envelope-from mmead@goof.com) Received: (qmail 30396 invoked by uid 10000); 23 Dec 2002 04:05:27 -0000 Date: Sun, 22 Dec 2002 23:05:27 -0500 From: "matthew c. mead" To: Matthew Dillon Cc: Robert Watson , freebsd-net@FreeBSD.ORG Subject: Re: TCP interactions (was: Re: Linux and FreeBSD poor network performance) Message-ID: <20021222230527.A30324@goof.com> References: <20021221165745.A67089@goof.com> <20021222151259.A21189@goof.com> <200212222106.gBML6INI032302@apollo.backplane.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: <200212222106.gBML6INI032302@apollo.backplane.com>; from dillon@apollo.backplane.com on Sun, Dec 22, 2002 at 01:06:18PM -0800 Sender: owner-freebsd-net@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org Hmm, same cables, same switch, different card, now it all works. I did do some ICMP ping testing with ping -f and only lost 3 packets after letting it run for a good 10s. -matt On Sun, Dec 22, 2002 at 01:06:18PM -0800, Matthew Dillon wrote: > By your description, it is almost certainly a packet loss problem... > a cabling issue or a switch issue most likely. Try doing large > pings, like this, and see if you get hicups: > > bsdbox# ping -i 0.1 -s 3000 linuxbox > > -Matt > Matthew Dillon > > > : > :On Sun, Dec 22, 2002 at 10:59:52AM -0500, Robert Watson wrote: > :> On Sat, 21 Dec 2002, matthew c. mead wrote: > : > :> > I have a Linux box and FreeBSD box sitting on a 100Mbit ethernet segment > :> > that cannot seem to talk to one another faster than 150K/s. I've been > :> > using scp, ftp, http, to test this. > : > :> And you've done tests in both directions, or just in one? > : > :Both. > : > :> Could you try using a non-TCP performance measurement tool of some sort? > :> I.e., some sort of UDP throughput test. Because of your comment about the > :> two boxes talking to windows fine, it sounds likely to be a TCP > :> interaction, but it would be useful to check and see. > : > :I didn't get around to it before I started looking at > :hardware/driver. I dropped a 3c905 into the Linux box and things > :improved. I grabbed a new ethernet card elsewhere today and it > :works just fine. > : > :> I've CC'd Matt Dillon because he's fixed a number of subtle TCP bugs of > :> this sort in the past and can probably provide some debugging guidance. > : > :I think it's the Linux driver. Sorry for the false alert. This would > :be a fun one to figure out if it were tcp stack interactions. :( > : > : > :-matt > : > :-- > :matthew c. mead > : > :http://www.goof.com/~mmead/ > -- matthew c. mead http://www.goof.com/~mmead/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message