[NCLUG] Need help with ether md raid or possibly ext4 corrupted group descriptors

[Sean Reifschneider]

Mike Jensen wrote:
> I did not think I needed a super block because it should use the ones that

Of course you needed a superblock.  The only time you don't need a
superblock is if you either are trying to access a (very) old array that
was created using a format before they had superblocks.  You didn't want to
do a build, you probably wanted to do an assemble:

       -A, --assemble
              Assemble a pre-existing array.

As far as why the boot disc didn't detect the array, I wonder if you had
the partitions set up as type "fd" in fdisk.  That means "RAID Autodetect",
the kernel will look in the partitions for md superblocks and assemble the
arrays it finds.

> array, and overwrite the superblocks thus forgetting the previous raid
> configuration.

Indeed.  As you say, overwrite the superblocks, meaning your existing array
had superblocks.  The hint that you might have been going the wrong
direction with --build is that it says that it operates on an array that
has no superblocks.  From the mdadm man page:

       This  usage  is similar to --create.  The difference is that it creates
       an array without a superblock. With these arrays there is no difference
       between  initially  creating  the array and subsequently assembling the
       array, except that hopefully there is useful data there in  the  second
       case.

> I think the reason the rescue cd could not find and mount the array is
> because /etc/mdadm.conf was contained on the array, so it could not

The rescue disc and the booting should not use /etc/mdadm.conf -- otherwise
you couldn't have /etc on the array.  They use the superblocks and the
partition-type of "fd" to assemble the array.

> remember how to rebuild the previous configuration.  If I told it to mount
> the drives it would say "no linux partitions found".

You can't just ask to mount a single device of a multi-disc array.  Mount
knows nothing about the array, and the array knows nothing about the
file-system.  Mount just works on block devices, and md just turns multiple
block devices into a single (larger and/or more redundant) block device.

With the exception of a RAID-1 array, directly mounting a component
partition of the array will probably mess things up pretty badly.  With a
RAID-1 array, you might be able to mount it read-only, but mounting it
read-write may result in file-system corruption or keeping only the
original data.

Sean
-- 
Sean Reifschneider, Member of Technical Staff <jafo at tummy.com>
tummy.com, ltd. - Linux Consulting since 1995: Ask me about High Availability

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 251 bytes
Desc: OpenPGP digital signature
URL: <http://lists.nclug.org/pipermail/nclug/attachments/20090131/de693294/attachment.pgp>