I mentioned in another thread that I needed to recover a RAID array when the RAID superblock on one of the partitions had become corrupted.
I could not find any documentation that deals with this - most of the examples assume you are hot swapping a failed device from an array that is running whereas in my case the system would not start the array concerned.
If I ran:
mdadm -A /dev/md0 /dev/sda7 /dev/sdb7 /dev/sdc7 /dev/sdd1
I would get an error message about the superblock on /dev/sda7. After much fiddling about the answer turned out to be to add the --force option after which the array is started without /dev/sda7. After that the corrupt but working partition can be added back in:
mdadm /dev/md0 -a /dev/sda7
and at that point the array is resynchronised to include /dev/sda7 .
Regards, Steve.