From: James Howells <james@howells.uk.net>
To: linux-raid@vger.kernel.org
Subject: Missing Drives
Date: Wed, 21 Jul 2010 01:29:41 +0100 [thread overview]
Message-ID: <201007210129.41566.james@howells.uk.net> (raw)
Hi everyone.
Fantastic job with the Linux mdadm tool but I've got a problem that I need
some support with.
(Before you all ask, I've Googled this one to death, so please don't think of
me as ignorant and blind).
Basically, I have a 5 disk RAID-5 array in an Addonics external enclosure.
Each disk is 1TB in size. Since I am running it over USB, it's takes 3-3.5
days to resync the disks if something goes pear-shaped.
Now, I don't think the software is at fault because I've had the drive
replaced with a brand new disk and it's doing the exact same thing -
drive /dev/sde loses it's partition data (the drive's superblock) and the md
device's superblock. The data is unaffected since I only ever mount the
device as read-only unless I need RW capability. I physically power the
device down after unmounting the partition and stopping the array.
Whilst other people have had this problem of drives suddenly "losing
everything", they have been able to recover, but I cannot keep my computer
running for 3-3.5 days just to rebuild good data!
My question is:
Is it possible to copy the md superblock from one known good drive to another
and assemble it as forced-clean?
It's easy enough to recreate the partition table with a simple:
dd if=/dev/sda of=/dev/sde count=1 bs=512
command but I believe the mdadm superblock lives at the end of the drive and
if I could recreate that then I could at least recover from a failure of a
known-good drive without needing a resync.
I'm looking into getting the RAID controller card (actually an eSATA 5-1 mini
card) replaced or repaired but in the meantime, I am exploring all avenues.
This only happens on /dev/sde and not on any other device. Swapping hard
drives around doesn't make any difference either - /dev/sde is perfectly
readable. Clearly, though, I am not happy about building an array with a
drive in that "cursed" slot since the loss of another drive in a RAID-5 would
destroy everything stored on the array.
Any advice would be gratefully appreciated.
Regards,
James
PS. I'm running Debian Lenny 2.6.26-2-686 32-bit kernel with mdadm version
2.6.7.2-3 (repackaged for Debian).
next reply other threads:[~2010-07-21 0:29 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-07-21 0:29 James Howells [this message]
2010-07-21 1:23 ` Missing Drives Mark Knecht
2010-07-21 2:06 ` Miles Fidelman
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=201007210129.41566.james@howells.uk.net \
--to=james@howells.uk.net \
--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;
as well as URLs for NNTP newsgroup(s).