From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alexander Junghans Subject: Raid 5 Superblock redundant? Date: Mon, 09 Sep 2002 14:26:17 +0200 Sender: linux-raid-owner@vger.kernel.org Message-ID: <3D7C9369.7010802@leukefeld.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: To: linux-raid@vger.kernel.org List-Id: linux-raid.ids Hi, first of all thanks for the Raidsupport in the Kernel : -> using it for Years! I have a general question about the implementation Raid 5. It happents me twice that one of a 3 Disk Raid 5 faild. After new installing the faild drive the system has to read the complete 2 disks to rebuild the 3. But there will be also a bad block, so the system will kick the 2. Disk and the Raid will fail. Now the Question: If I would give you 3 Blocks of a Raid 5 same Position differend Disk could you tell me which is worg? A + B +C = Data Simpy fake A (A) + B + C =Data Without knowing the superblock that a has failed. The idea behind this ist to make a raid 5 readable with some faid Blocks on any drive with the fact that no more than 1 Block on each "slide" is down. Block 1 A + B + C = O.k. Block 2 (A) +B + C = O.k. ... Block 8 A + (B) + C = O.k. ------------------------------- All Data O.K. I would love to have a mode of Raid 5 where no drive is kicked and all data are O.K, just for saving the Data. Would this tecnially possible? Thanks.