From mboxrd@z Thu Jan 1 00:00:00 1970 From: Patrick McHardy Subject: Re: cups slow on linux-2.6.24 Date: Wed, 30 Jan 2008 14:47:22 +0100 Message-ID: <47A07FEA.3000702@trash.net> References: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15; format=flowed Content-Transfer-Encoding: 7bit Cc: Jozsef Kadlecsik , lkml , Krzysztof Piotr Oledzki , "David S. Miller" , cups-bugs , Netfilter Development Mailinglist To: Jeff Chua Return-path: Received: from stinky.trash.net ([213.144.137.162]:51601 "EHLO stinky.trash.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752145AbYA3Nri (ORCPT ); Wed, 30 Jan 2008 08:47:38 -0500 In-Reply-To: Sender: netfilter-devel-owner@vger.kernel.org List-ID: Jeff Chua wrote: > On Jan 29, 2008 6:53 PM, Jozsef Kadlecsik wrote: > >> As the problem can be reproduced so easily, could you capture a full TCP >> session and send the pcap file? Thus it could be analyzed, replayed, etc. >> and found the reason why the patch above slows down the printing. >> By reverting back it we loose TCP reopening support. > > Not an expert here. I use 'tcpdump -i lo' and don't see anything > different between the good and bad. Anyway, attached are the good and > bad log files. > > There's a slight pause of 27 seconds after 502 jobs (both good and bad > runs), then the good one will resume and transmit in burst until > completing all the 1000 times. The bad run will resume and transmit > one-by-one every 3 seconds. > > Please tell me what options to use with tcpdump to dump more useful output. A binary dump would be more useful: tcpdump -i lo -w and I guess Jozsef also wants "-s 0" so the full packets are included.