Hi, I’ve found myself in possession of a QNAP that needs deleted files recovered from it. I am having trouble activating the MD array in Ubuntu so I can run ext4magic on it.
The array is an MD RAID10 of 14 disks with LVM2/EXT4 on top (+ an SSD cache). So far I have:
No dice. It doesn’t activate all of the arrays. I also notice that the list of Personalities is much shorter (sorry no paste, I didn’t do it remotely).
Any pointers?
I may try to compile ext4magic in QTS instead of using Ubuntu…
This is just a crapshoot of ancient experience. But last time I remember messing with mdadm when you scan it will build the conf according to the raid metadata on the disks.
So it should reassemble an existing array. I dont know how this will fair if a drive is missing.
The fact that he nabbed the config off the qnap box should keep it from rebuilding from disk metadata.
I’ve scratched my head for a solid 20 minutes, and the best I can come up with is “I need more information from the ubuntu box”. If you get around to having an SSH session that you can copy and paste from, feel free to share more details.
The only thing that immediately comes to mind would be, “Is it possible the array did assemble, and the volume group wasn’t activated?”
Ok, back to not all md arrays working. For the main array (md256), Ubuntu thinks it’s made up of 12 spares and 0 disks… idk what to do with that. It’s a 14 drive RAID 10.
I would absolutely do that if it was viable. The array is about 30TB. I can’t imagine how long that would take to dd, and there’s no hardware for it anyway.
It’s a RAID10 of 14 6TB drives. The LV is 30tb. We left room for some overhead. QNAP uses a snapshot for rsync backups so you want some play there and it’s always a good idea to have some extra space…
Update: Apparently, reconstructing QNAP’s md/lvm stack is legitimately difficult. I just got a $65k quote for a 4-6 week timeline only because they couldn’t sort out the md/lvm which is completely functional and fine in the QTS OS.
At this point, it looks like dd’ing the whole EXT4 fs to an image file on another NAS is the most viable option… that sucks though.
Have you tried checking the event count on each of the partitions mdadm -E /dev/sdf2 /dev/sdo2 ... and then trying to run an assemble with just the disks that have matching event counts?
Additionally, I am here waiting for a DHL courier delivery of 8 drives from Newegg that is supposed to arrive by midnight. These drives need to be added to the destination NAS before the transfer reaches 10TB.