Originally Posted by koflanagan
Hmm 3 out of 7 disk are dead in a raid 6.. I would say is not recoverable..
No actually only one disk is really dead. Two more howeveer are somehow out-of-date. One was removed from the array and re-added as a new disk, but the recovery-process stopped at 2% due to a Controller failure. Another disk was also removed from the RAID and then re-added, but it was taken up as a "spare" disk, not re-integrated in the RAID, although nothing on the disk was changed (it was part of the array before and just got kicked out due to the same Controller failure).
So basically I started with a RAID with 7 disks, then
* one disk died and was removed.
* RAID had 6 drives left
* An unknown error kicked disk A out
* RAID had 5 drives left
* Disk A was checked with SMART and came up error-free
* Disk A was added to the RAID
* Array started "recovery" on Disk A and aborted at 2%
* After reboot, Disk A and Disk B was missing from the Array
* YUK - only 4 Drives in the Array
* Did mdadm --add on Disk B, effectively adding it as "spare"
* YUK now it's 4 drives and 1 spare although Disk B was the 5th Disk!
* Controller was determined to be the source of the problem
* Made a dd-Copy of all HDDs to experiment with recovery options
So Disk B should still contain the data, unless adding it as spare (however that happened) deleted all the content.
And Disk A should at least contain 98% of the data that was in there before the recovery started.
It is not vital to recover all of the content, it would be ok to recover most of it.