Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S263231AbVCEAUe (ORCPT ); Fri, 4 Mar 2005 19:20:34 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S263421AbVCEASa (ORCPT ); Fri, 4 Mar 2005 19:18:30 -0500 Received: from smtp06.auna.com ([62.81.186.16]:54176 "EHLO smtp06.retemail.es") by vger.kernel.org with ESMTP id S263316AbVCDXzf convert rfc822-to-8bit (ORCPT ); Fri, 4 Mar 2005 18:55:35 -0500 Date: Fri, 04 Mar 2005 23:55:21 +0000 From: "J.A. Magallon" Subject: Re: Something is broken with SATA RAID ? To: Jeff Garzik Cc: Lista Linux-Kernel References: <1109810381l.5754l.0l@werewolf.able.es> <20050303005210.GA1140@havoc.gtf.org> In-Reply-To: <20050303005210.GA1140@havoc.gtf.org> (from jgarzik@pobox.com on Thu Mar 3 01:52:10 2005) X-Mailer: Balsa 2.3.0 Message-Id: <1109980521l.13844l.0l@werewolf.able.es> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 3337 Lines: 93 On 03.03, Jeff Garzik wrote: > On Thu, Mar 03, 2005 at 12:39:41AM +0000, J.A. Magallon wrote: > > Hi... > > > > I posted this in other mail, but now I can confirm this. > > > > I have a box with a SATA RAID-5, and with 2.6.11-rc3-mm2+libata-dev1 > > works like a charm as a samba server, I dropped it 12Gb from an > > osx client, and people does backups from W2k boxes and everything was fine. > > With 2.6.11-rc4-mm1, it hangs shortly after the mac starts copying > > files. No oops, no messages... It even hanged on a local copy (wget), > > so I will discard samba as the buggy piece in the puzzle. > > > > I'm going to make a definitive test with rc5-mm1 vs rc5-mm1+libata-dev1. > > I already know that plain rc5-mm1 hangs. I have to wait the md reconstruction > > of the 1.2 TB to check rc5-mm1+libata (and no user putting things there...) > > Please eliminate -mm and -libata-dev from the equation. > One piece at last... I have tried - 2.6.11 - 2.6.11 + libata-dev1 + netdev1 + shrinkers-at-tail + 1Gb-lowmem Bot work fine and survived several gigas dumped both through smb and afp. Happy man ;). If there was something strange, it must be in -mm. rc5-mm1 did not work, but plain 2.6.11 works. I will try 2.6.11-mm1 on monday... Just a note. Net throughput seems a bit slower in the second case (measured with iftop). And it degrades over time. With a 8 Gb copy, it started at about 50Mb/s and dropped to 25 at the end. Not sure if the one to blame is linux or osx... Hardware (just for the record): 2 x PDC20319 (FastTrak S150 TX4) (rev 02) 6 x 250Gb Maxtor 7Y250M0 YAR5, 3 on each. RAID-5 nada:~# mdadm -D /dev/md0 /dev/md0: Version : 00.90.01 Creation Time : Fri Sep 3 02:17:28 2004 Raid Level : raid5 Array Size : 1225557760 (1168.78 GiB 1254.97 GB) Device Size : 245111552 (233.76 GiB 250.99 GB) Raid Devices : 6 Total Devices : 6 Preferred Minor : 0 Persistence : Superblock is persistent Update Time : Fri Mar 4 21:34:00 2005 State : clean Active Devices : 6 Working Devices : 6 Failed Devices : 0 Spare Devices : 0 Layout : left-symmetric Chunk Size : 256K UUID : fd6fcad0:21da140b:072a82b1:11b3db21 Events : 0.156336 Number Major Minor RaidDevice State 0 8 1 0 active sync /dev/sda1 1 8 17 1 active sync /dev/sdb1 2 8 33 2 active sync /dev/sdc1 3 8 49 3 active sync /dev/sdd1 4 8 65 4 active sync /dev/sde1 5 8 81 5 active sync /dev/sdf1 /dev/md0: Timing cached reads: 756 MB in 2.01 seconds = 376.93 MB/sec Timing buffered disk reads: 158 MB in 3.02 seconds = 52.26 MB/sec Thanks. -- J.A. Magallon \ Software is like sex: werewolf!able!es \ It's better when it's free Mandrakelinux release 10.2 (Cooker) for i586 Linux 2.6.11-jam1 (gcc 3.4.3 (Mandrakelinux 10.2 3.4.3-3mdk)) #1 - 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/