Failed raid 01 - mdadm

[yagerd001]

Horrors. I setup a raid using /dev/sda and /dev/sdb. Within a couple of weeks, sdb failed with a repetitive clicking sound. I immediately turned off my computer and shipped the thing back to Seagate, and they sent me a replacement. Now when I print the partition table for sda using fdisk, it says:

Warning: invalid flag 0x0000 of partition table 4 will be corrected by w(rite)

Evidently I didn't follow the procedure to "fail" the bad drive, and now I can't figure out how to retrieve the information on the good drive, /dev/sda.

Anyone help me out of this jam?

Oddly enough, the replacement disc shows that it's already partitioned.

DY

email:
yagerd001@hawaii.rr.com

You should be able to fix this in four steps:


first guest partition to see what it thought was really there by issuing this command

eqgpart /dev/hdc

Secondly save the MBR and the partition table

dd if=/dev/hdc of=mbr bs=512 count=1

Third, use gpart and tell it that you have faith in what it
believed was in the first partition, and for it to re-create the
partition table

gpart -W /dev/hdc /dev/hdc

Fourth reboot after running gpart.

This should help

[yagerd001]

Gparted shows that /dev/sda is unallocated. It does give me the option of assigning a disk label. If I assign a disk label, will that damage any contents that still might be there?

[Flyboy917]

I dissagree on this. There are specific tools/ in md or mdadm used to rebuild a failed drive in an array. Unfortunatly, I can't tell you what they are because its been such a long since I set up my raid. When I did originally set it up, I intentionally "failed" a drive just to test my ability to restore it. It was simple a simple command line if I recall. Be carefull with gpart.

here is a start:
http://tldp.org/HOWTO/Software-RAID-0.4x-HOWTO-4.html