From mboxrd@z Thu Jan 1 00:00:00 1970 From: Phil Turmel Subject: Re: Raid5 drive fail during grow and no backup Date: Fri, 07 Nov 2014 22:36:26 -0500 Message-ID: <545D8FBA.9090701@turmel.org> References: <5455A35C.2060000@turmel.org> <5458FC2A.1050308@turmel.org> <545CEDFB.6060806@gautschi.net> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <545CEDFB.6060806@gautschi.net> Sender: linux-raid-owner@vger.kernel.org To: "P. Gautschi" Cc: Vince , linux-raid@vger.kernel.org List-Id: linux-raid.ids On 11/07/2014 11:06 AM, P. Gautschi wrote: > > This is a problem you haven't solved yet, I think. The raid array > should have fixed this bad sector for you without kicking the drive out. > The scenario is common with "green" drives and/or consumer-grade drives > in general. > > ... > > Then you can set up your array to properly correct bad sectors, and > set your system to look for bad sectors on > > a regular basis. > > What is the behavior of mdadm when a disk reports a read error? > - reconstruct the data, deliver it to the fs and otherwise ignore it? > - set the disk to fail? > - reconstruct the data, rewrite the failed data and continue with any > action? > - rewrite the failed data and reread it (bypassing the cache on the HD)? Option 3. Reconstruct and rewrite. However, if the device with the bad sector is trying to recover longer than the linux low level driver's timeout, bad things^TM happen. Specifically, the driver resets the SATA (or SCSI) connection and attempts to reconnect. During this brief time, it will not accept further I/O, so the write back of the reconstructed data fails. Then the device has experienced a *write* error, so MD fails the drive. This is the out-of-the-box behavior of consumer-grade drives in raid arrays. > Do read operation always read the parity too in order to detect problems > early > before a sector on a other disks fails? No. > Can the behavior be configured in any way? I found no documentation > regarding this. The administrator must schedule "check" scrubs of the array to look for bad sectors, or wait for them to be found naturally. Such scrubs will also find inconsistent parity and report it. A "repair" scrub can then fix the broken parity. I understand that some distros include a cron job for this purpose. I've always rolled my own. Phil