From owner-freebsd-stable Sun Oct 12 16:42:37 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id QAA28372 for stable-outgoing; Sun, 12 Oct 1997 16:42:37 -0700 (PDT) (envelope-from owner-freebsd-stable) Received: from awfulhak.demon.co.uk (awfulhak.demon.co.uk [158.152.17.1]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id QAA28367 for ; Sun, 12 Oct 1997 16:42:29 -0700 (PDT) (envelope-from brian@awfulhak.org) Received: from gate.lan.awfulhak.org (localhost [127.0.0.1]) by awfulhak.demon.co.uk (8.8.5/8.8.5) with ESMTP id AAA10331; Mon, 13 Oct 1997 00:38:52 +0100 (BST) Message-Id: <199710122338.AAA10331@awfulhak.demon.co.uk> X-Mailer: exmh version 2.0zeta 7/24/97 To: dg@root.com cc: John Dowdal , Jerry Hicks , stable@FreeBSD.ORG Subject: Re: ftp transfers crawling... In-reply-to: Your message of "Sun, 12 Oct 1997 09:47:36 PDT." <199710121647.JAA03859@implode.root.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 13 Oct 1997 00:38:52 +0100 From: Brian Somers Sender: owner-freebsd-stable@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > >> Since about Wednesday's make world of 2.2-STABLE, I can't make *any* ftp > >> download do better that 1K on my 28.8k modem . That's trying a lot of > >> different sites. I'm using IJPPP. All was well before, no config > >> changes have been made. > >> > >> Maybe Netcom... Anybody else notice anything? > > > >I am having similar problems. Few reproducable ones: This has prompted me to do some speed testing.... the results look good - probably because I'm comparing against a rather outdated pppd-2.2.0 (from FreeBSD 2.2.2). All tests are transfering a 1Mb file that's all zeros accross a null-modem ppp link: 3.0 current rel 2.2.2 PRED1 VJ Time Oct 12 ppp ppp N Y 1:33 ppp ppp N N 1:35 ppp ppp Y Y 0:12 ppp pppd-2.2.0 N Y 3:17 pppd-2.3.1 pppd-2.2.0 N N 3:17 pppd-2.3.1 ppp N Y 1:33 Of course the PRED1 case is useless given the file contents (but it's nice to see that it does something). The VJ results are disappointing, although given that I have a transmission unit of 1500 and the slot id isn't compressed talking ppp<->ppp (I'll enable this in -current soon, but it's too dangerous for 2.2.2 at this point), VJ is reducing 40+1460 bytes down to about 4+1460 - a 2.5% gain is close to the above (actual) 2.1% gain. Pppd-2.2.0 just doesn't seem to compare with anything. The second line is the most interesting (comparing rather exactly to the last line). Serial line throughput (excluding all the ACK and setup crud) is 88301 bps. With all the crud, this 1Mb turns into 1092851 bytes sent and 29000 bytes received. 1092851 bytes over 95 seconds = 92030 bps (assuming the UART *is* actually syncronous). I may try to make these tests a bit more informing - I need a test bed (read: set of sanity checks) so that I don't screw things up again like I just did. -- Brian , , Don't _EVER_ lose your sense of humour....