From: Wols Lists <antlists@youngman.org.uk>
To: bruce.korb+reply@gmail.com
Cc: linux-raid@vger.kernel.org
Subject: Re: Trying to rescue a RAID-1 array
Date: Thu, 31 Mar 2022 22:34:17 +0100 [thread overview]
Message-ID: <8c2148d0-fa97-d0ef-10cc-11f79d7da5e5@youngman.org.uk> (raw)
In-Reply-To: <CAKRnqNLjsX9nVLrLedo4tfxtg0ZBz=6XJu=-z_Ebw6Auh+oz-Q@mail.gmail.com>
On 31/03/2022 19:14, Bruce Korb wrote:
> On Thu, Mar 31, 2022 at 10:06 AM Wols Lists <antlists@youngman.org.uk> wrote:
>>
>> On 31/03/2022 17:44, Bruce Korb wrote:
>>> I moved the two disks from a cleanly shut down system that could not
>>> reboot and could not
>>> be upgraded to a new OS release. So, I put them in.a new box and did an install.
>>> The installation recognized them as a RAID and decided that the
>>> partitions needed a
>>> new superblock of type RAID-0.
>>
>> That's worrying, did it really write a superblock?
>
> Yep. That worried me, too. I did the command to show the RAID status of the two
> partitions and, sure enough, both partitions were now listed as RAID0.
>
>>> Since these data have never been
>>> remounted since the
>>> shutdown on the original machine, I am hoping I can change the RAID
>>> type and mount it
>>> so as to recover my. .ssh and .thunderbird (email) directories. The
>>> bulk of the data are
>>> backed up (assuming no issues with the full backup of my critical
>>> data), but rebuilding
>>> and redistributing the .ssh directory would be a particular nuisance.
>>>
>>> SO: what are my options? I can't find any advice on how to tell mdadm
>>> that the RAID-0 partitions
>>> really are RAID-1 partitions. Last gasp might be to "mdadm --create"
>>> the RAID-1 again, but there's
>>> a lot of advice out there saying that it really is the last gasp
>>> before giving up. :)
>>>
>>
>> https://raid.wiki.kernel.org/index.php/Asking_for_help
>
> Sorry about that. I have two systems: the one I'm typing on and the one
> I am trying to bring up. At the moment, I'm in single user mode building
> out a new /home file system. mdadm --create is 15% done after an hour :(.
> It'll be mid/late afternoon before /home is rebuilt, mounted and I'll be
> able to run display commands on the "old" RAID1 (or 0) partitions.
>
>> Especially lsdrv. That tells us a LOT about your system.
>
> Expect email in about 6 hours or so. :) But openSUSE doesn't know
> about any "lsdrv" command. "cat /proc/mdstat" shows /dev/md1 (the
> RAID device I'm fretting over) to be active, raid-0 using /dev/sdc1 and sde1.
Well, the webpage does tell you where to download it from - it's not
part of the official tools, and it's a personal thing that's damn useful.
>
>> What was the filesystem on your raid? Hopefully it's as simple as moving
>> the "start of partition", breaking the raid completely, and you can just
>> mount the filesystem.
>
> I *think* it was EXT4, but. it might be the XFS one. I think I let it default
> and openSUSE appears to prefer the XFS file system for RAID devices.
> Definitely one of those two. I built it close to a decade ago, so I'll be moving
> the data to the new /home array.
>
>> What really worries me is how and why it both recognised it as a raid,
>> then thought it needed to be converted to raid-0. That just sounds wrong
>> on so many levels. Did you let it mess with your superblocks? I hope you
>> said "don't touch those drives"?
>
> In retrospect, I ought to have left the drives unplugged until the install was
> done. The installer saw that they were RAID so it RAID-ed them. Only it
> seems to have decided on type 0 over type 1. I wasn't attentive because
> I've upgraded Linux so many times and it was "just done correctly" without
> having to give it a lot of thought. "If only" I'd thought to back up
> email and ssh.
> (1.5TB of photos are likely okay.)
>
> Thank you so much for your reply and potentially help :)
>
If it says the drive is active ...
When you get and run lsdrv, see if it finds a filesystem on the raid-0 -
I suspect it might!
There's a bug, which should be well fixed, but it might have bitten you.
It breaks raid arrays. But if the drive is active, it might well mount,
and you will be running a degraded mirror. Mount it read-only, back it
up, and then see whether you can force-assemble the two bits back
together :-)
But don't do anything if you have any trouble whatsoever mounting and
backing up.
Cheers,
Wol
next prev parent reply other threads:[~2022-03-31 21:34 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-03-31 16:44 Trying to rescue a RAID-1 array Bruce Korb
2022-03-31 17:06 ` Wols Lists
2022-03-31 18:14 ` Bruce Korb
2022-03-31 21:34 ` Wols Lists [this message]
2022-04-01 17:59 ` Bruce Korb
2022-04-01 18:21 ` Bruce Korb
2022-04-01 19:45 ` Wol
2022-04-01 20:23 ` Bruce Korb
2022-04-01 21:02 ` Wol
2022-04-01 21:24 ` Bruce Korb
2022-04-01 21:33 ` Wol
2022-04-20 8:40 ` Need to move RAID1 with mounted partition Leslie Rhorer
2022-04-20 8:55 ` Roman Mamedov
2022-04-20 9:21 ` Leslie Rhorer
2022-04-20 11:08 ` Andy Smith
2022-04-20 12:07 ` Pascal Hambourg
2022-04-20 12:24 ` Leslie Rhorer
2022-04-20 12:16 ` Leslie Rhorer
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=8c2148d0-fa97-d0ef-10cc-11f79d7da5e5@youngman.org.uk \
--to=antlists@youngman.org.uk \
--cc=bruce.korb+reply@gmail.com \
--cc=linux-raid@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox