Software mdadm RAID6 recovery - how to reassemble badly broken array
Hello,
I had set up a 2tb drive x 8 RAID6 array (12tb total size) on an external box linked to my server by 2 esata cables. 1 big 2tb GPT partition on each disk marked as RAID autodetect. The whole thing was formatted as an ext4 12tb file system.
Yesterday I noticed that the array was down with 4 disks showing as being out of the array (perhaps half of the array linked to one of the esata cables had disappeared briefly).
After rebooting the machine all 8 disks showed ok status on smartctl and the raid autodetect partition on each disk showed up ok in gdisk.
I tried the various mdadm reassemble options (including force) with no luck. I should have also tried with assumeclean but did not.
After reading some articles I tried mdadm --create --level 6 --raid-devices 8 with various chunk sizes, since I did not remember what chunk size I had used. Each time, some resyncing of the array took place. Each time, trying to mount the md device with mount -t ext4 /dev/mdx failed with no superblock error.
I have a few questions:
* Is the actual data on the disks gone due to the mdadm resyncing or is it all still intact?
* If the data is still there, is there some way of determining what the original raid chunk size was, by looking at the actual data on the disk?
* Once I get the chunk size figured out, and do an mdadm create, is there some way I can force the mount to work, or some other way to piece together the data on the various disks in order to recover it?
Thanks so much in advance,
--Ed
|