From owner-freebsd-bugs Mon Nov 27 19:23:11 2000 Delivered-To: freebsd-bugs@freebsd.org Received: from k6.office.htslabs.com (adsl-63-202-243-139.dsl.sndg02.pacbell.net [63.202.243.139]) by hub.freebsd.org (Postfix) with ESMTP id 7F2CE37B479 for ; Mon, 27 Nov 2000 19:23:08 -0800 (PST) Received: from htslabs.com (localhost [127.0.0.1]) by k6.office.htslabs.com (8.11.1/8.11.1) with ESMTP id eAS3MO500635; Mon, 27 Nov 2000 19:22:25 -0800 (PST) (envelope-from bsd_admin@htslabs.com) Message-ID: <3A2324F0.3AE50202@htslabs.com> Date: Mon, 27 Nov 2000 19:22:24 -0800 From: William Schmidt Organization: Thomson Instrument Co. X-Mailer: Mozilla 4.73 [en] (X11; U; FreeBSD 4.2-RELEASE i386) X-Accept-Language: en MIME-Version: 1.0 To: David Greenman Cc: freebsd-bugs@FreeBSD.ORG Subject: Re: kern/22768: fxp get slow often! References: <200011130620.WAA26438@freefall.freebsd.org> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-bugs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org David Greenman wrote: > The following reply was made to PR kern/22768; it has been noted by GNATS. > > From: David Greenman > To: Wei-Kai Wu > Cc: FreeBSD-gnats-submit@FreeBSD.ORG > Subject: Re: kern/22768: fxp get slow often! > Date: Sun, 12 Nov 2000 22:10:51 -0800 > > >On Sat, Nov 11, 2000 at 07:36:52AM -0800, David Greenman wrote: > >> >ifconfig fxp0 140.113.214.152 netmask 0xffffff00 media 100BaseTX > >> >sleep 1 > >> >ifconfig fxp0 140.113.214.152 netmask 0xffffff00 media autoselect > >> >running 2 times every hour to keep the speed of ftpd. > >> I have no idea what you are trying to accomplish with the above, but it is > >> weird in the extreme. The problem that you are describing sounds like a duplex > >> negotiation problem with your switch. Try setting both sides to full duplex. > >> For the FreeBSD side remove your cronjob and then: > >> ifconfig .... media 100basetx mediaopt full-duplex > > > >I am quite sure that the Intel 8-port 10/100 Switch work well. > >I also tried your suggestion, still not work. > > > >In fact, I used the dc0 network card before. > >I thought the problem may cause by dc0, so I buy the fxp0. > > > >Is there any other possibilities? > > So you're having the same problem with 'tulip' clones (dc device)? Have you > tried using a different cable? Does the Intel switch have any error stats that > might give a clue why the performance is bad? What does "ifconfig fxp0" show > (speed/duplex) when the performance is low? > > -DG I have been work on this problem for sum time now and have determined that It seems to be the data coming in that is slow. For example I can ftp put at about 2.7-3.0 MB/sec but only ftp get the same file at about 40-120 KB/sec. and that was manually setting ifconfig for 100baseTX and full-duplex. Manually setting to 10BaseT results at a rate of 700-900KB/secin both directions. Furthermore If I reboot the same system to 3.5.1-release the problem goes away and rate in both direction are over 2.7MB/sec. I have taken all my systems back to 3.5.1 to temporally solve these problems and left one dual boot system for testing. Bill To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-bugs" in the body of the message