Debian RAID 1 Array Crashed; Trying to mount Disk 2

[mpovah]

Greetings!

I am a freelancer just brought in today on this problem and I'm stumped.

My client had their email server running on a Debian install with software Raid 1 across two identical 40 gig Maxtor IDE hard drives. The OS was on disk 1, and disk 2 was used as the mirror.

The other day disk 1 failed. Disk 2 has no boot sector it seems, or is unmountable for some other reason. It won't boot from disk 2, anyways, and I was unable to mount it. Fsck says "DriveReady SeekComplete Error" and that it cannot read sector 0 (I ran "fsck /dev/hdb") and it acts as if there is no /hdb1, 2 etc.

I understand that Raid 1 is disk mirroring - is the data on disk 2? Can it possibly be accessed? If so, how?

Thanks for any help!

- Matt

[Mikey]

Did you go back to basics and check to see if the bios reads the second drive? Ctrl-a at the scsi bios as well?

Then make sure the boot disk your using is going for the kernal in the right mirrored boot disk. IE if it was hda1 change it to hdb1. You could always just image a new drive from the mirror and mout it as hda1 and let it take care of itself. Faster that way.

Where in NS are you?

[mpovah]

Hi,

Thanks for the reply.

Well the drive is definitely being read by the BIOS. There is no SCSI (the original RAID was with two IDE Maxtor drives.)

Newbie alert - can you give me more details as to the specifics of the steps you recommended? The array won't mirror on its' own unfortunately - it won't boot. Apparently the boot image was on Disk 1, and the mirror was just the data partitions, possibly not even including the boot partition! Don't ask me why they did it that way... (Sad.) Oh well.

Thanks,

Matt

P.S. Southwest Nova

[Giro]

Get a live CD boot from that and mount it like any other disk.

[mpovah]

Hi Giro,

What I did was I connected it as a slave drive in a Red Hat 9 box I had, and tried to mount the drive normally, but I got the error messages as related in my first post, the drive wouldn't mount (I used "mount -t /dev/hdb (etc.) /<mount directory> and it would say "read error sector 0" or something like that.)

Perhaps I just don't know what the specification should be for the partition number, but I used hdb, hdb1, 2, 3, 4, etc. Tried up to seven.

This is software RAID 1 that apparently came with a Debian distro the previous (now long gone) admin had set up, perhaps he set it up incorrectly?

Any help greatly appreciated, thanks,

Matt

[Giro]

Maybe this could help what have you tried apart from a normal mount?

[mpovah]

Hi Giro,

Well I tried the standard mount procedure as outlined previously, and then I tried fsck, it gave another error. Finally I tried the file system debugger but I couldn't open the drive with that either.

That's about all I knew to try. The mirror disk (Disk 2) which is still operational won't boot, and it acts like there are no partitions on it, I had hoped that I'd be able to mount it normally on my RH9 box and recover the data, but no dice.

Thanks,

Matt

[Mikey]

If the boot wasn't mirrored. My suggestion on imaging or ghosting the drives won't work. Can you boot using a windows boot disk, put fdisk on it, and then check the two disks and see if they have identical partitions? ie did the boot partition get mirrored as well.

So ultimately, you need to recreate a boot partition. I wish I could help if that's your direction, but I don't think I can.

BTW I lived in Lake Echo, just outside Dartmouth. Now in Hamilton Ontario