From owner-freebsd-net@FreeBSD.ORG Thu Oct 8 17:05:37 2009 Return-Path: Delivered-To: freebsd-net@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 1DE201065670; Thu, 8 Oct 2009 17:05:37 +0000 (UTC) (envelope-from rihad@mail.ru) Received: from mx74.mail.ru (mx74.mail.ru [94.100.176.89]) by mx1.freebsd.org (Postfix) with ESMTP id AAA878FC27; Thu, 8 Oct 2009 17:05:36 +0000 (UTC) Received: from [217.25.27.27] (port=52432 helo=[217.25.27.27]) by mx74.mail.ru with asmtp id 1MvwQo-0004p8-00; Thu, 08 Oct 2009 21:05:34 +0400 Message-ID: <4ACE1B95.2090508@mail.ru> Date: Thu, 08 Oct 2009 22:04:21 +0500 From: rihad User-Agent: Mozilla-Thunderbird 2.0.0.22 (X11/20090706) MIME-Version: 1.0 To: Julian Elischer References: <4AC9E29B.6080908@mail.ru> <20091005123230.GA64167@onelab2.iet.unipi.it> <4AC9EFDF.4080302@mail.ru> <4ACA2CC6.70201@elischer.org> <4ACAFF2A.1000206@mail.ru> <4ACB0C22.4000008@mail.ru> <20091006100726.GA26426@svzserv.kemerovo.su> <4ACB42D2.2070909@mail.ru> <20091006142152.GA42350@svzserv.kemerovo.su> <4ACB6223.1000709@mail.ru> <20091006161240.GA49940@svzserv.kemerovo.su> <4ACC5563.602@mail.ru> <4ACC56A6.1030808@mail.ru> <4ACC5DEC.1010006@mail.ru> <4ACC65A0.7030900@mail.ru> <4ACC8CC8.8050403@mail.ru> <4ACDE1FB.6020602@mail.ru> <4ACE171D.9080707@elischer.org> In-Reply-To: <4ACE171D.9080707@elischer.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Spam: Not detected X-Mras: Ok Cc: Eugene Grosbein , Robert Watson , Luigi Rizzo , freebsd-net@freebsd.org Subject: Re: dummynet dropping too many packets X-BeenThere: freebsd-net@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Networking and TCP/IP with FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 08 Oct 2009 17:05:37 -0000 Julian Elischer wrote: > you can not do anything about it if one of the custommers sends a burst > of 3000 udp packets at their maximum speed(or maybe some combination of > custommers to something which results in an aggreagate burst rate like > that. > In other words you may always continue to get moments when the pipe > releases a bunch of stuff that has a potential to over-run something > down stream. > I've said this before, but: the PC is only dealing with the downstream traffic, i.e. traffic arriving from the net. I won't say anything against dummynet bursts overfilling hardware buffers, but they're _only_taking place when the number of entries in the ipfw tables reaches 2000 or so, and the traffic load is at about 430-450 mbps. That is, it _never_ happens before both conditions are true. Although raising HZ from 1000 up to 4000 hasn't helped, which is somewhat contrary to the idea of bursts. My idea is to switch to an Intel 10GigE card, which is unavoidable sooner or later anyway, given our natural growth. Probably an upgrade to 8.0 will do some good, too. > Think of it as a dam in a stream... > > If you have no dam, teh water level goes up and down gradually and by > small amounts, but if you have a dam, you can release water in such a > way that the stream is flooded higher than it would normally ever get. > > > work out how large a burst of data the pipe will release in 1/4000th of > a second and using a small packet size, work out how many packets that > is. Then make sure that the driver software input queue is > capable of holding that many packets. > >