From mboxrd@z Thu Jan 1 00:00:00 1970 From: Simon Horman Subject: Re: Possible regression in HTB Date: Thu, 9 Oct 2008 22:11:02 +1100 Message-ID: <20081009111100.GA28667@verge.net.au> References: <48EB5A92.6010704@trash.net> <20081007220022.GA2664@ami.dom.local> <20081008002153.GL12021@verge.net.au> <48EBFF5E.1090902@trash.net> <48EC0190.7040804@trash.net> <48EC6286.1030202@cdi.cz> <20081008085325.GF4174@ff.dom.local> <48EC8FDD.5030507@cdi.cz> <20081009010957.GB6342@verge.net.au> <48EDA339.3040706@cdi.cz> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Jarek Poplawski , Patrick McHardy , netdev@vger.kernel.org, David Miller To: Martin Devera Return-path: Received: from kirsty.vergenet.net ([202.4.237.240]:51747 "EHLO kirsty.vergenet.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754736AbYJILLF (ORCPT ); Thu, 9 Oct 2008 07:11:05 -0400 Content-Disposition: inline In-Reply-To: <48EDA339.3040706@cdi.cz> Sender: netdev-owner@vger.kernel.org List-ID: On Thu, Oct 09, 2008 at 08:22:49AM +0200, Martin Devera wrote: > >> > >> Simon, can you try to these things (separately): > >> a/ increase quantum to the first class (say 10x) > > > > Hi Martin, > > > > Do you mean increase r2q ? If so, here are some results > > no, no, add rather ... "quantum 50000" to the first class only > r2q affects computation for all classes Sorry, I wasn't aware of that parameter. I ran a test now. With the default quantum (6250) for the 500Mbit/s class I got: 10194: 383384949bits/s 383Mbits/s 10197: 285879669bits/s 285Mbits/s 10196: 287648424bits/s 287Mbits/s ----------------------------------- total: 956913042bits/s 956Mbits/s And when a quantum of 50000 for the 500Mbit/s class I got: 10194: 396345613bits/s 396Mbits/s 10197: 279511994bits/s 279Mbits/s 10196: 281062498bits/s 281Mbits/s ----------------------------------- total: 956920106bits/s 956Mbits/s > >> b/ set ceil=rate on all three classes > > > > This seems pretty close to the expected/ideal result. > > > > 10194: 496575074bits/s 496Mbits/s > > 10197: 96969861bits/s 96Mbits/s > > 10196: 96973002bits/s 96Mbits/s > > ----------------------------------- > > total: 690517938bits/s 690Mbits/s > > > > Hmm, it seems that problem is with drr loop or borrowing, > rate computation seems to work ok even in this corner case. > > In any case from your other test follows that when NIC > throttles then HTB can't arrive at correct rates - I feel > this as HTB bug and add it to my (very long) todo list... > > thanks, Martin -- Simon Horman VA Linux Systems Japan K.K., Sydney, Australia Satellite Office H: www.vergenet.net/~horms/ W: www.valinux.co.jp/en