QNAP RAID 5 recovery after a power outage: healthy drives, lost structure

All four 1TB drives in a QNAP TS-410 RAID 5 array were showing as unmounted, the result of a power outage. The disks also showed a different firmware version to the NAS box.

All four drives copied with 0 bad sectors, so unreadable sectors were ruled out. The disk order wasn't correct: it was 1, 4, 3, 2 and not the expected 1, 2, 3, 4. The file system structures appeared to have changed, and the metadata was no longer present.

With the metadata gone, the array's layout had to be worked out from the disks. The engineers found a RAID configuration that could be used, and scanned from there for the best possible result.

1. Removed the PCBs from all four drives and cleaned the contacts.
2. Copied all four drives sector by sector, with 0 bad sectors.
3. Checked the disk order and found it had changed.
4. Worked out a usable set of RAID parameters, then scanned for the best possible result.
5. Extracted the data to a 2TB portable drive.

1.47TB of data came back, a partial recovery. The root folder names were lost, but the subfolder names and their contents were fine. RAW files, recovered without their names, are included in that total.

Healthy drives alone don't make an array readable: here every sector read, and the layout was still the problem. The disk order and metadata are part of the data's structure, so when they're lost, the file system has to be worked out again and the result can be partial. That the QNAP had been re-initialised made the job harder. This was a logical failure: the hardware worked, and the structure around the data was what had been lost.

What changed the disk order. That the same result is possible on every re-initialised NAS. How long the work took, or what it cost.

1. Healthy drives on their own don't make an array readable; the layout and metadata can be what's lost.
2. Don't re-initialise the NAS or re-create the volume to "get it back". That writes new array metadata over the old layout.