Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S275262AbTHGJxy (ORCPT ); Thu, 7 Aug 2003 05:53:54 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S275263AbTHGJxx (ORCPT ); Thu, 7 Aug 2003 05:53:53 -0400 Received: from smtp017.mail.yahoo.com ([216.136.174.114]:56324 "HELO smtp017.mail.yahoo.com") by vger.kernel.org with SMTP id S275262AbTHGJxX convert rfc822-to-8bit (ORCPT ); Thu, 7 Aug 2003 05:53:23 -0400 From: Michael Buesch To: Nuno Silva Subject: Re: [2.6] system is very slow during disk access Date: Thu, 7 Aug 2003 11:53:02 +0200 User-Agent: KMail/1.5.3 References: <200308062052.10752.fsdeveloper@yahoo.de> <200308062247.17816.fsdeveloper@yahoo.de> <3F31D205.2080008@vgertech.com> In-Reply-To: <3F31D205.2080008@vgertech.com> Cc: Jean-Yves LENHOF , insecure@mail.od.ua, linux kernel mailing list , linux-ide@vger.kernel.org, Andrew Morton MIME-Version: 1.0 Content-Type: Text/Plain; charset=US-ASCII Content-Transfer-Encoding: 7BIT Content-Description: clearsigned data Content-Disposition: inline Message-Id: <200308071153.18361.fsdeveloper@yahoo.de> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2138 Lines: 84 -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Thursday 07 August 2003 06:13, Nuno Silva wrote: > Hi! Hi, > There are some references to: hdparm -a 512 /dev/hda. > For some folks this improves fs performance. > > Does it work for you? Changing read-ahead doesn't work. Neither to 8 nor to 512. I think it doesn't have something to do with read-ahead, because the system-performance-drop occurs while writing to the disk. My dd-example wrote to disk: dd if=/dev/zero of=./t.test I just ran a few other tests: dd if=/dev/hda of=/dev/null doesn't drop system performance and the system is very usable while dd is running. So reading from a plain partition is OK. dd if=/dev/zero of=/dev/hdc7 This slows down the system _very_ much, but it's still usable. The mouse curser doesn't jump randomly over the screen, like it did while dd if=/dev/zero of=./t.test So my guess, it may have something to do with reiserFS. So I ran another test: I did mke2fs /dev/hdc7 mount /dev/hdc7 /mnt/data_1 cd /mnt/data_1 dd if=/dev/zero of=./t.test And now the surprise. :) Here the system-behaviour is exactly the same, as while writing to the plain partition. The cursor is not smooth, but it's usable and it doesn't jump ramdonly over the screen. So I think it has something to do with reiserFS. Where can I start to track it down, why reiserFS is doing this? Hmm, I'll try to enable CONFIG_REISERFS_CHECK and CONFIG_REISERFS_PROC_INFO Can somebody tell me how to debug reiserFS with these options enabled? > Regards, > Nuno Silva Short note: hda and hdc are both exactly the same devices. - -- Regards Michael Buesch [ http://www.8ung.at/tuxsoft ] Penguin on this machine: Linux 2.6.0-test2 - i386 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.1 (GNU/Linux) iD8DBQE/MiGKoxoigfggmSgRAt+PAJ9gI8Z5Osm4/DcnFZsPtR7x4UQpZACfZw7R GxC2PwCVOAWFFE39Fw9ttKM= =v0nD -----END PGP SIGNATURE----- - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/