From owner-freebsd-questions@FreeBSD.ORG Wed Jun 25 03:29:14 2003 Return-Path: Delivered-To: freebsd-questions@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 198CB37B401 for ; Wed, 25 Jun 2003 03:29:14 -0700 (PDT) Received: from luke.cpl.net (luke.cpl.net [216.117.199.213]) by mx1.FreeBSD.org (Postfix) with ESMTP id 6A0F543FF3 for ; Wed, 25 Jun 2003 03:29:13 -0700 (PDT) (envelope-from shawn@cpl.net) Received: from shawn (ip-216-117-219-84.keyway.net [216.117.219.84]) by luke.cpl.net (8.11.3/8.11.3) with SMTP id h5PATAV41365; Wed, 25 Jun 2003 03:29:10 -0700 (PDT) Message-ID: <007a01c33b04$70dfd0d0$54db75d8@shawn> From: "Shawn Ramsey" To: "Karl Pielorz" References: <009701c339ed$b89daf40$85dd75d8@shawn> <512328439.1056443294@Study.tdx.com> <20030624170155.A46097@cpl.net> <57377156.1056532581@raptor> Date: Wed, 25 Jun 2003 03:27:55 -0700 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.50.4807.1700 X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4910.0300 cc: freebsd-questions@freebsd.org Subject: Re: Network Performace X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 25 Jun 2003 10:29:14 -0000 > That's certainly a step in the right direction :) > > > But something is still very > > wrong performance wise. It has helped, but I still can't push in/out > > nearly 100Mb/sec. (100Mb in, 100Mb out I mean). A simple FTP transfer > > locally through the routers gigabit interface causes our internet > > performance to plummet. I've disabled all the onboard stuff that was > > sharing IRQs with PCI cards, but I didn't figure that was an issue, > > didn't make a difference either way. Would the fact the gigabit is on the > > same PCI bus have any bearing? I would expect to at least get 100BT > > performance even so, but I don't have any experience with gigabit > > ethernet... > > The only thing I can suggest is try different PCI slots, or Gigabit cards, > or, worst case a different system. Having NIC's on separate PCI busses (as > opposed to both on the same PCI bus) may help it [But that's probably going > to need a new board etc.] Thats what I was going to try next, but i've noticed this : gw# netstat -I xl0 -w 1 input (xl0) output packets errs bytes packets errs bytes colls 7120 11 9118416 5566 0 2507132 0 6754 6 8498898 5649 0 3009719 0 7104 9 8856812 5802 0 3004529 0 7737 19 9787501 6429 0 2994557 0 7551 16 9670490 5948 0 2761794 0 I would guess such a large number of errors per second could be causing the interrupt usuage? From what i've been able to find such input errors are usually caused by a bad cable, or switch... > You don't say what Gigabit nic's your using? - I've had a lot of varied > results with different nic's, with surprisingly cheap 10/100/1000Mbit cards > giving 'reasonable' performance - but get left standing for dust by other > more expensive cards. Its an Intel Fiber card, don't know the exact model offhand... > The only other thing I can think of is, check the duplex/media options are > all setup properly on the cards / switches etc. - or try forcing things to > fdx etc. Been there, done that, full duplex/autoselect makes no difference, but its currently set for full/1000SX on both ends.