Hello Folks,

I'm having problem in re-creating RAID-1 array.

Here is the background to my problem.
(Sorry it's bit long)
I had two scsi disks with raid-1 setup on them.
I replaced them with new ones.

Before adding to RAID-1 array I formatted them.
mkfs.ext3 /dev/sdb
mkfs.ext3 /dev/sdc

After which I created partitions on them.(intermediate commands excluded)
'fdisk /dev/sdb' to create /dev/sdb1.
'fdisk /dev/sdc' to create /dev/sdc1.

After removing old scsi disks, I had accidentally deleted /dev/md0.
So I had to recreate it once again.

mdadm --create -a /dev/md0 --level=raid1 --raid-devices=2 /dev/sdb1 /dev/sdc1
'-a' option created /dev/md0 for me.

Now the problem is
#mdadm --query /dev/md0
Line1 excluded(Sorry I'm not sitting on the same machine)
/dev/md0: No md super block found, not an md component.

#mdadm --examine /dev/md0
mdadm: No super block found on /dev/md0

And the strange thing is RAID-1 is working fine.
'cat /proc/mdstat' gives out normal messages.

So when I restart the system, mounting of /dev/md0 fails saying super block number is invalid and it is expecting ext2 file system.

The matter is quite urgent so any help is welcome.

Thank you,
Ravinandan

Did you create a filesystem on /dev/md0?

mkfs -t ext3 /dev/md0

... or something...

No..I have not run that command.

One silly doubt.
Should I run this command before adding the scsi disk to array or after adding the scsi disks.

There is already data written on this disks.
I stopped the RAID array and then tried to run this command.
Here is what I got.

#umount /dev/md0
#mdadm -S

#mkfs -t ext3 /dev/md0
mkfs.ext3: Device size reported to be zero. Invalid partition specified, or partition table wasn't reread after running fdisk, due to a modified partition being busy and in use. You may need to reboot to re-read your partition table.

Please suggest how should I proceed...

Ravinandan

There is already data written on this disks.
... that's a problem. mkfs will overwrite that data. Good thing it didn't work.

#mkfs -t ext3 /dev/md0
mkfs.ext3: Device size reported to be zero
... looks like you need to add the drives to md0.

# cat /proc/mdstat
... explicitly remove anything that shouldn't be there. (mdadmin -r)

This is just md0 of a RAID1 that is being rebuilt right? You want to mirror partitions on the existing drive to the new one and their geometery is identical and everything?

So the trick is to try adding the mirror partitions like normal:

$ sudo mdadm -a /dev/md0 /dev/sda1
$ sudo mdadm -a /dev/md0 /dev/sdb1
$ sudo mdadm -a /dev/md1 /dev/sdb2
... etc

http://www.heronforge.net/redhat/node14.html

>So the trick is to try adding the mirror partitions like normal:
This is where the problem started.
After removing the failed disk with a new one, I added it to the raid array.
Unfortunately the resync(mirroring) failed at 34% due to I/O read error from the old scsi disk.
resync was restarted automatically but kept failing at 34%.

So I just copied all the files from old scsi disk thru 'cp' command in preserve mode.
Removed the old scsi disk and added a new one.

Now I recreated the RAID-1 array.
mdadm --create -a /dev/md0 --level=raid1 --raid-devices=2 /dev/sdb1 /dev/sdc1

The sync was successful.

But the only problem is I can't reboot since /dev/md0 has different magic number.

I'm going to rebuild the RAID from scratch I hope that will solve the problem.

Ravinandan

Yep... that'll do it.