From owner-freebsd-questions Tue Dec 17 1:47:35 2002 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 65DD237B401 for ; Tue, 17 Dec 2002 01:47:31 -0800 (PST) Received: from nemesis.systems.pipex.net (nemesis.systems.pipex.net [62.241.160.8]) by mx1.FreeBSD.org (Postfix) with ESMTP id 914CE43EA9 for ; Tue, 17 Dec 2002 01:47:30 -0800 (PST) (envelope-from stacey@vickiandstacey.com) Received: from [192.168.1.8] (81-86-129-77.dsl.pipex.com [81.86.129.77]) by nemesis.systems.pipex.net (Postfix) with ESMTP id EBA95160073B9; Tue, 17 Dec 2002 09:47:13 +0000 (GMT) Subject: Re: Upload substantially slower than download From: Stacey Roberts Reply-To: stacey@vickiandstacey.com To: Mark Cc: FreeBSD Questions In-Reply-To: <200212170831.GBH8VKM42847@asarian-host.net> References: <200212170027.GBH0RUW29189@asarian-host.net> <1040085387.86677.105.camel@asarian-host.net> <1040109376.86677.125.camel@localhost> <200212170831.GBH8VKM42847@asarian-host.net> Content-Type: text/plain Organization: Message-Id: <1040118446.86677.157.camel@localhost> Mime-Version: 1.0 X-Mailer: Ximian Evolution 1.2.0 Date: 17 Dec 2002 09:47:27 +0000 Content-Transfer-Encoding: 7bit Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Hi Mark, Thanks for the output. If its an ftp transfer then please try specify the port number (21) and re-run the test transfers. One thing that has shown itself is the difference in window sizes depending on the direction of the transfer: UPLOAD (WinXP to FBSD?) 09:22:13.470561 test-server.net..ssh > 192.168.79.3.1845: . ack 3064209 win > 56940 (DF) < window size in use DOWNLOAD (FreeBSD to WinXP?) 09:25:53.111169 192.168.79.3.1845 > test-server.net..ssh: . ack 7843933 win > 64240 (DF) test-server.net..ssh: P 3060397:3061289(892) ack 992 win 63760 (DF) Dload: ?09:25:53.109400 192.168.79.3.1845 > test-server.net..ssh: . ack 7842077 win 64240 (DF) To verify: test-server.net is the FBSD box 192.168.79.3 is the WinXP box On establishing an Uload connection during the (WInXP to FBSD) the max window size is set to 63760, and for a Dload connection, this is declared as 64240 - in practice, this need not, and isn't the case here. Before leaping to uncertain conclusions at this point, I would ask that you run tcpdump with options: tcpdump ip host test-server.net and 192.168.79.3 The reason for my hesitation here, is the fact that that I would prefer for someone more knowledgeable of FBSD's tcp control algorithms to (hopefully shed some light on whether or not FBSD's stack throttles throughput depending on direction. This is something usually handled at the link layer (nic), but might actually be configurable at the kernel level with sysctl. In the mean time, see if its possible from WinXP to hard-set the MTU or max window size at that end. Regards, Stacey On Tue, 2002-12-17 at 08:31, Mark wrote: > ----- Original Message ----- > From: "Stacey Roberts" > To: "Mark" > Cc: "FreeBSD Questions" > Sent: Tuesday, December 17, 2002 8:16 AM > Subject: Re: Upload substantially slower than download > > > > Do the transfers (in both directions) with tcpdump running on the > > FBSD box. You would get the complete conversation between > > both boxes, and be able to see what is actually happening at > > the packet level. > > > Hi Tracey (and others), > > Thanks for the continued help. Ok, I did a tcpdump; like this: > > "tcpdump host test-server.net and port 22" > > This is what I get back on an UPLOAD: > > -------------------------------------------------------------- > 09:22:13.371011 192.168.79.3.1845 > test-server.net..ssh: P > 3060397:3061289(892) ack 992 win 63760 (DF) > 09:22:13.439905 test-server.net..ssh > 192.168.79.3.1762: . ack 11857 win > 58400 (DF) [tos 0x10] > 09:22:13.469905 test-server.net..ssh > 192.168.79.3.1845: . ack 3061289 win > 58400 (DF) > 09:22:13.470415 192.168.79.3.1845 > test-server.net..ssh: . > 3061289:3062749(1460) ack 992 win 63760 (DF) > 09:22:13.470539 192.168.79.3.1845 > test-server.net..ssh: . > 3062749:3064209(1460) ack 992 win 63760 (DF) > 09:22:13.470561 test-server.net..ssh > 192.168.79.3.1845: . ack 3064209 win > 56940 (DF) > 09:22:13.470662 192.168.79.3.1845 > test-server.net..ssh: . > 3064209:3065669(1460) ack 992 win 63760 (DF) > 09:22:13.470685 test-server.net..ssh > 192.168.79.3.1845: . ack 3065669 win > 58400 (DF) > 09:22:13.470785 192.168.79.3.1845 > test-server.net..ssh: . > 3065669:3067129(1460) ack 992 win 63760 (DF) > 09:22:13.470907 192.168.79.3.1845 > test-server.net..ssh: . > 3067129:3068589(1460) ack 992 win 63760 (DF) > 09:22:13.470923 test-server.net..ssh > 192.168.79.3.1845: . ack 3068589 win > 56940 (DF) > 09:22:13.470984 192.168.79.3.1845 > test-server.net..ssh: P > 3068589:3069481(892) ack 992 win 63760 (DF) > 09:22:13.569901 test-server.net..ssh > 192.168.79.3.1845: . ack 3069481 win > 58400 (DF) > 09:22:13.570477 192.168.79.3.1845 > test-server.net..ssh: . > 3069481:3070941(1460) ack 992 win 63760 (DF) > 09:22:13.570599 192.168.79.3.1845 > test-server.net..ssh: . > 3070941:3072401(1460) ack 992 win 63760 (DF) > 09:22:13.570618 test-server.net..ssh > 192.168.79.3.1845: . ack 3072401 win > 56940 (DF) > 09:22:13.570722 192.168.79.3.1845 > test-server.net..ssh: . > 3072401:3073861(1460) ack 992 win 63760^C > > 6614 packets received by filter > 0 packets dropped by kernel > -------------------------------------------------------------- > > And this on a DOWNLOAD (which goes real fast): > > -------------------------------------------------------------- > 09:25:53.109400 192.168.79.3.1845 > test-server.net..ssh: . ack 7842077 win > 64240 (DF) > 09:25:53.110732 test-server.net..ssh > 192.168.79.3.1845: . > 7842473:7843933(1460) ack 13128 win 58400 (DF) > 09:25:53.110762 test-server.net..ssh > 192.168.79.3.1845: . > 7843933:7845393(1460) ack 13128 win 58400 (DF) > 09:25:53.110781 test-server.net..ssh > 192.168.79.3.1845: . > 7845393:7846853(1460) ack 13128 win 58400 (DF) > 09:25:53.110793 test-server.net..ssh > 192.168.79.3.1845: . > 7846853:7848313(1460) ack 13128 win 58400 (DF) > 09:25:53.110881 test-server.net..ssh > 192.168.79.3.1845: . > 7848313:7849773(1460) ack 13128 win 58400 (DF) > 09:25:53.111169 192.168.79.3.1845 > test-server.net..ssh: . ack 7843933 win > 64240 (DF) > 09:25:53.111346 192.168.79.3.1845 > test-server.net..ssh: . ack 7846853 win > 64240 (DF) > 09:25:53.111565 192.168.79.3.1845 > test-server.net..ssh: . ack 7849773 win > 64240 (DF) > 09:25:53.111584 test-server.net..ssh > 192.168.79.3.1845: P > 7849773:7850153(380) ack 13128 win 58400 (DF) > 09:25:53.112766 test-server.net..ssh > 192.168.79.3.1845: . > 7850153:7851613(1460) ack 13128 win 58400 (DF) > 09:25:53.112789 test-server.net..ssh > 192.168.79.3.1845: . > 7851613:7853073(1460) ack 13128 win 58400 (DF) > 09:25:53.112821 test-server.net..ssh > 192.168.79.3.1845: . > 7853073:7854533(1460) ack 13128 win 584^C > -------------------------------------------------------------- > > I cannot say these numbers mean a great deal to me (yet); does this kind of > output make any sense? > > Thanks again! > > - Mark > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-questions" in the body of the message -- Stacey Roberts B.Sc (HONS) Computer Science Web: www.vickiandstacey.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message