From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1161071AbWFKDRD (ORCPT ); Sat, 10 Jun 2006 23:17:03 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1161074AbWFKDRD (ORCPT ); Sat, 10 Jun 2006 23:17:03 -0400 Received: from trinity.phys.uwm.edu ([129.89.57.159]:16584 "EHLO trinity.phys.uwm.edu") by vger.kernel.org with ESMTP id S1161071AbWFKDRB (ORCPT ); Sat, 10 Jun 2006 23:17:01 -0400 Date: Sat, 10 Jun 2006 22:15:59 -0500 (CDT) From: Bruce Allen X-X-Sender: ballen@trinity.phys.uwm.edu To: Justin Piszcz cc: Jan-Benedict Glaw , apiszcz@solarrain.com, smartmontools-support@lists.sourceforge.net, Remy Card , linux-kernel@vger.kernel.org, "Theodore Ts'o" , David Beattie Subject: Re: [smartmontools-support] The Death and Diagnosis of a Dying Hard Drive - Is S.M.A.R.T. useful? In-Reply-To: Message-ID: References: <20060610105141.GE30775@lug-owl.de> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Justin, It's an unfortunate fact of life that SMART will not detect all disk failures. My research group in the U. Wisconsin - Milwaukee Physics Department runs two large computing clusters (approximately 2000 hard disks total). We run weekly extended self-tests with smartd. Our experience over about five years is that about 2/3 of drive failures can are predicted by smartd. The other 1/3 of failures have no warning. I am surprised that the extended self-test does not detect the bad sectors on your disk. Our experience is that the typical SYSLOG 'seek failure' error messages do correlate very well with the failing LBAs found via SMART self-tests. On your disk, it may be the case that these bad sectors are *sometimes* readable, or that the sequential scanning done during a SMART self-test do not provoke these errors. If you have some time to follow up, you could do some experiments with a recent release of dd using the 'direct' option to bypass the block layers in the Linux kernel. Cheers, Bruce