From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Jarosch Subject: Re: TCP connection stalls under 2.6.24.7 Date: Fri, 25 Jul 2008 17:34:26 +0200 Message-ID: <200807251734.27743.thomas.jarosch@intra2net.com> References: <47EA0DAB.7080205@securenet.de> <200807251500.05527.thomas.jarosch@intra2net.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-15 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Jozsef Kadlecsik , Netdev , Patrick McHardy , Sven Riedel , Netfilter Developer Mailing List , "=?iso-8859-15?q?D=E2niel?= Fraga" , David Miller To: "Ilpo =?iso-8859-15?q?J=E4rvinen?=" Return-path: Received: from rs02.intra2net.com ([81.169.173.116]:39566 "EHLO rs02.intra2net.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750807AbYGYPec convert rfc822-to-8bit (ORCPT ); Fri, 25 Jul 2008 11:34:32 -0400 In-Reply-To: Content-Disposition: inline Sender: netdev-owner@vger.kernel.org List-ID: Ilpo, On Friday, 25. July 2008 16:06:04 Ilpo J=E4rvinen wrote: > I guess the patch below would be enough in itself to create the desir= ed > effect (though "desired" is hardly a negative enough word to describe= a > workaround of this kind).=20 Yeah, the result feels good enough for me. Here's the latest tcpdump before I run out of good filenames for the dumps: http://www.intra2net.com/de/download/tcpdump/tcp_frto_combined_patch.ta= r.bz2 > Ie., in the worst case 6.8% of your link's capacity was wasted during= the > transfer due to inefficiency cause by that middlebox, not counting th= e > under-utilization that occurs both because of a small window or a wai= t for > RTOs, not bad result at all... :-D IIRC our outbound box does traffic shaping, so some percents are to be=20 accounted to packets being dropped to slow down the connection a bit=20 if they come (in) too fast. Anyway, we now just have to flip a coin if this gets into the kernel or not :-) I really hope this could save someone from doing the same debug session all over again... Thanks for the hard work you put into debugging this. Thomas