From owner-freebsd-stable@FreeBSD.ORG Tue May 31 01:22:42 2005 Return-Path: X-Original-To: freebsd-stable@freebsd.org Delivered-To: freebsd-stable@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id E086816A41C for ; Tue, 31 May 2005 01:22:42 +0000 (GMT) (envelope-from vinny@tellurian.com) Received: from mail1.tellurian.net (mail1.tellurian.net [216.182.1.23]) by mx1.FreeBSD.org (Postfix) with ESMTP id 79DF343D1D for ; Tue, 31 May 2005 01:22:42 +0000 (GMT) (envelope-from vinny@tellurian.com) Received: from leviathon.tellurian.com (leviathon.tellurian.net [216.182.41.250]) by mail1.tellurian.net ([216.182.1.23] Tellurian Networks Mail Server version 3.0d-2) with ESMTP id 227729776 for multiple; Mon, 30 May 2005 21:22:41 -0400 Message-Id: <6.2.3.0.2.20050530211857.040ff6a0@pop3.tellurian.com> X-Mailer: QUALCOMM Windows Eudora Version 6.2.3.0 Date: Mon, 30 May 2005 21:22:47 -0400 To: Sebastian Ahndorf From: Vinny Abello In-Reply-To: <429B7DF6.8040704@it-is-warlock.de> References: <20050529221024.4fu2p4yjusk04k0g@mail.banot.net> <20050529212705.GA64753@xor.obsecurity.org> <1117447400.5384.4.camel@localhost.localdomain> <20050530104928.GB79877@sr.se> <1117465224.9934.7.camel@localhost.localdomain> <429B44BD.7070806@pp.nic.fi> <20050530191843.GA82875@xor.obsecurity.org> <429B7DF6.8040704@it-is-warlock.de> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed X-Authenticated-User: vinny@tellurian.com X-Ultimate-Internet-Connection: Tellurian Networks Cc: Imobach ??? Sosa , freebsd-stable@freebsd.org, Kris Kennaway Subject: Re: Poor network performance: a lot of timeouts X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 31 May 2005 01:22:43 -0000 At 04:56 PM 5/30/2005, Sebastian Ahndorf wrote: >Kris Kennaway wrote: >>>Both sides must have same config, autosense should work if there >>>is no config possibility in other end. >> >>autosense may in fact not work, especially on low-quality NICs like rl. > >I don't agree to that. >I had similar problems with my network using a cheap switch with >some realtek nics. I had the nics running 100baseTX Full Duplex. >Changing this to autosense made the problems gone. > >Reason (as some people of the german questions-list told me): >Many cheap switches always send their autosensepakets, and have >great problems if the nics connected to the switch do not response >to the autosensepakets (cause they are configured to 10/100baseTX >full/half duplex). >Also realtek nics are far away from being good nics, they work >without problems with the autosensemode and a cheap switch for me >(and many other people I know). > >I would suggest the starter of this thread to use autosense with his >nic (if not tested yet). The deal is simply this: Autosense must be enabled on both sides to autonegotiate speed/duplex. If you force one side to full duplex, the other side still autosenses (on most unmanaged switches), fail, and will fall back to half duplex causing a duplex mismatch. You have to force the other side to full duplex as well. If you cannot do this, leave it at auto, or set the side you can manage to half so they agree. I personally like to leave everything on auto except links between switches and between routers and switches which I force to full. It always works out well for me on practically any platform or OS. Maybe on one or two occasions I've experienced faulty drivers which cause the autosensing to not work on the NIC and just default to half duplex. Vinny Abello Network Engineer Server Management vinny@tellurian.com (973)300-9211 x 125 (973)940-6125 (Direct) PGP Key Fingerprint: 3BC5 9A48 FC78 03D3 82E0 E935 5325 FBCB 0100 977A Tellurian Networks - The Ultimate Internet Connection http://www.tellurian.com (888)TELLURIAN "Courage is resistance to fear, mastery of fear - not absence of fear" -- Mark Twain