I take a look at what happened with my Buffalo NAS, and how to recover data from its dead corpse.
A few years ago, when everything was getting more and more connected, I decided to buy a NAS. It shouldn’t be expensive, so I got a Buffalo LS-WXL. It had two 2TB drives, and some seriously shitty hardware specs, but it was enough at that time. Not beeing completely stupid, I set it up to back up the first drive to the second one every week (which I thanked myself for later).
But then, two months ago, it started to email me write errors on one of the hard drives (the one I backed up to). I thought “A broken hard drive, especially a desktop HDD, that can happen” and wanted to replace it. That was when the NAS turned off (or hung, whatever). I got it back online, and copied all the stuff to another drive. The next morning, the NAS was off again. And it didn’t turn on properly again, so I took it off the network, removed the drives and attached them to a PC.
It was fairly easy to copy all the data off since it wasn’t striped (RAID 0) (There is a great guide on how to do that at skelleton.net, and you should probably read it first anyway). The file system Buffalo uses is XFS, the proper drivers are included in xfs-progs on Linux. When looking at the partition layout of the drive with lsblk, you will see six partitions on the drive, with the last one having a Linux MD RAID 1 in it. On the right, the name of the array, for example md123 is shown. You should then be able to then mount read-only (very much recommended!) it with mount -o ro /dev/md123 /mnt/hdd.
You can then copy it off using either rsync or just plain ol’ cp like this:
cp -av /mnt/hdd/* /backuplocation
The a flag will preserve any modification dates on files and the v makes if verbose (shows what file is currently getting copied).
So the data was safe, but what happened? When doing SMART-Tests (there are many guides on how to do this, I used GSmartControl to make it easy) of the first drive, everything was good:
Num Test_Description Status Remaining LifeTime(hours) LBA_of_first_error
# 1 Short offline Completed without error 00% 19206 -
Other values to look out for are Reallocated_Sector_Ct, Offline_Uncorrectable and Current_Pending_Sector. These will tell you how many bad sectors were reallocated, how many can not be reallocated and how many are still pending to be reallocated. If you get a few on the first value, that is mostly fine. Sectors do die. For the first drive, everything was okay again:
5 Reallocated_Sector_Ct 0x0033 100 100 036 Pre-fail Always - 0
197 Current_Pending_Sector 0x0012 100 100 000 Old_age Always - 0
198 Offline_Uncorrectable 0x0010 100 100 000 Old_age Offline - 0
Another value that caught my attention was this one:
190 Airflow_Temperature_Cel 0x0022 071 043 045 Old_age Always In_the_past 29 (0 4 29 27 0)
The hard drive got too hot in the NAS, which can significantly reduce it’s lifespan.
But the other drive (the one I backed up to) was way more interesting:
Num Test_Description Status Remaining LifeTime(hours) LBA_of_first_error
# 1 Extended offline Completed: read failure 90% 19205 62888117
# 2 Short offline Completed: read failure 90% 19205 62888117
That doesn’t look good at all. If we take a look at the previously mentioned values:
5 Reallocated_Sector_Ct 0x0033 051 051 036 Pre-fail Always - 64800
197 Current_Pending_Sector 0x0012 100 100 000 Old_age Always - 96
198 Offline_Uncorrectable 0x0010 100 100 000 Old_age Offline - 96
There are tutorials on how to fix single uncorrectable sectors by zeroing them out with dd, but 96 are way too many. Seeing the 64800 reallocated sectors, the hard drive probably ran out of spare sectors. I decided to crap that drive.
But what happened to the NAS itself? The truth is, I don’t know. Maybe it didn’t want to boot with all the drive errors, but considering that it doesn’t run SMART tests, I would rather bet on it crapping it’s pants at the same time as the hard drive.