From mboxrd@z Thu Jan 1 00:00:00 1970 From: Wols Lists Subject: Re: Two raid5 arrays are inactive and have changed UUIDs Date: Tue, 14 Jan 2020 15:23:40 +0000 Message-ID: <5E1DDCFC.1080105@youngman.org.uk> References: <959ca414-0c97-2e8d-7715-a7cb75790fcd@youngman.org.uk> <5E17D999.5010309@youngman.org.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-raid-owner@vger.kernel.org To: William Morgan Cc: linux-raid@vger.kernel.org List-Id: linux-raid.ids On 14/01/20 14:47, William Morgan wrote: > Well, I went ahead and tried the forced assembly: > > bill@bill-desk:~$ sudo mdadm --assemble --force /dev/md1 /dev/sdg1 > /dev/sdh1 /dev/sdi1 > [sudo] password for bill: > mdadm: Merging with already-assembled /dev/md/1 This looks like your problem ... it looks like you have a failed assembly active. Did you do an "mdadm --stop /dev/md1"? You should always do that between every attempt. > mdadm: Marking array /dev/md/1 as 'clean' > mdadm: failed to RUN_ARRAY /dev/md/1: Input/output error > I hope that's a good sign - if it's got an array that doesn't make sense then everything should have stopped at that point and the --force won't have done any damage. > (The drive letters have changed because I removed a bunch of other > drives. The original drives are now on sd[b,c,d,e] and the copies are > on sd[f,g,h,i] with sdf being a copy of the presumably bad sdb with > the event count which doesn't agree with the other 3 disks.) Make sure md1 doesn't appear to exist at all (--stop), and then try again ... Cheers, Wol