From owner-freebsd-ipfw@FreeBSD.ORG Tue Oct 14 12:51:20 2003 Return-Path: Delivered-To: freebsd-ipfw@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 0C23216A4B3 for ; Tue, 14 Oct 2003 12:51:20 -0700 (PDT) Received: from boole.cs.uh.edu (Boole.cs.uh.edu [129.7.240.11]) by mx1.FreeBSD.org (Postfix) with ESMTP id 48D8B43FD7 for ; Tue, 14 Oct 2003 12:51:19 -0700 (PDT) (envelope-from mzu@cs.uh.edu) Received: from mail.cs.uh.edu (pascal [129.7.240.15]) by boole.cs.uh.edu (Postfix) with SMTP id C9317F97C for ; Tue, 14 Oct 2003 14:51:18 -0500 (CDT) Received: from 63.172.179.2 (SquirrelMail authenticated user mzu) by mail.cs.uh.edu with HTTP; Tue, 14 Oct 2003 14:51:18 -0500 (CDT) Message-ID: <16380.63.172.179.2.1066161078.squirrel@mail.cs.uh.edu> Date: Tue, 14 Oct 2003 14:51:18 -0500 (CDT) From: mzu@cs.uh.edu To: freebsd-ipfw@freebsd.org User-Agent: SquirrelMail/1.4.1 MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 Importance: Normal Subject: HELP!!! Dummynet Causing Machines Dead?? X-BeenThere: freebsd-ipfw@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: IPFW Technical Discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 14 Oct 2003 19:51:20 -0000 I have three Squid machines working together as a cache array. I used Dummynet to control the inter-cache bandwidth. My three squid is 10.12.0.1, 10.12.0.2 and 10.12.0.3 I used the following commands to create pipes in each squid For example, in Squid1: ipfw add pipe 1 IP from 10.12.0.1 to 10.12.0.2 out ipfw add pipe 2 IP from 10.12.0.1 to 10.12.0.3 out ipfw pipe 1 config bw 10Mbit/s queue 75Kbytes ipfw pipe 2 config bw 10Mbit/s queue 75Kbytes I have another program to change the bandwidth of each pipe every half an hour, with a lot of traffic going through the caches at the same time. So that the configeration for a pipe can change FROM: ipfw pipe 1 config bw 10Mbit/s queue 73Kbytes TO: ipfw pipe 1 config bw 2Mbit/s queue 25Kbytes Then in very short time, the machine is dead. I think the problem might because of the setting of "queue". But I don't know how much should be correct. I was using "bw x Mbit/s queue 6*x+13 Kbytes". Oh, when bandwidth change from 5M to 3M ,or 8M to 5M (smaller difference), the machines will run just fine with my 6*x+13 setting. Only steep jump like from 8M to 2M will cause death. Please help me out!!! This is emergency!!!!