From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S266257AbUGOSA2 (ORCPT ); Thu, 15 Jul 2004 14:00:28 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S266263AbUGOSA2 (ORCPT ); Thu, 15 Jul 2004 14:00:28 -0400 Received: from zero.aec.at ([193.170.194.10]:41994 "EHLO zero.aec.at") by vger.kernel.org with ESMTP id S266257AbUGOSA1 (ORCPT ); Thu, 15 Jul 2004 14:00:27 -0400 To: Maciej Soltysiak cc: linux-kernel@vger.kernel.org Subject: Re: tcp_window_scaling degrades performance References: <2igbK-82L-13@gated-at.bofh.it> From: Andi Kleen Date: Thu, 15 Jul 2004 20:00:08 +0200 In-Reply-To: <2igbK-82L-13@gated-at.bofh.it> (Maciej Soltysiak's message of "Thu, 15 Jul 2004 17:20:08 +0200") Message-ID: User-Agent: Gnus/5.110003 (No Gnus v0.3) Emacs/21.2 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Maciej Soltysiak writes: > I have been experiencied weird problems with network throughput > lately and I after experimenting with /proc/sys/net/ipv4 knobs > I found that when I have tcp_window_scaling 0 I can > get throughput from a distant server of about 600kB/s (well, 200kB/s > is fast enough) It's pretty easy for you to find out. Do a tcpdump -v or ethereal -v from both the side of a host you download from and from the linux side. Then compare all packets. If they don't match the firewall is doing something bad. Especially check window values and TCP options in the SYN packets It is very very likely the firewall, window scaling works for a lot of people. -Andi