2010
01.06

I had purchased a 2TB enclosure to do offsite backups with at the beginning of last year, but kept putting off purchasing the new backup system to go with it, as the drives in my computers were largely full (my storage needs had reached the point that the solutions were $$$$).  As is the rule, exactly a week before I was to finally purchase said solution (I had been waiting for Black Friday hard drive prices), the enclosure died, taking everything I’d done that year with it (although I did have most of the creative things I’d done in more than one place).  If anyone should know better, I should know better, I managed the backups for a large-ish hospital for years, but I was cocky enough to think that things would keep working until I got the new solution in place.  Take this as a warning.  If you know better, you know better and need to back your stuff up.  If you don’t know better:

IF YOU DON’T HAVE YOUR IMPORTANT DATA IN MORE THAN ONE PLACE IT IS ONLY A MATTER OF TIME BEFORE ITS INEVITABLE LOSS.  IT IS NOT A QUESTION OF IF YOU WILL LOSE IT, BUT ONLY A QUESTION OF WHEN.

While some of the fancier enclosures run a version of linux, this particular enclosure, and many of the ones like it that contain two hard drives just use a JMicron (or similar) chipset to treat both of the drives as one drive, writing the first part of the partition on one of them and the last half on the second.  This is called spanning or, occasionally, BIG mode.  My first thought had been that mine had died because one of the two drives was dead, so was surprised when I ran dd_rescue (apparently the version with the _ is supposed to be better than the one without it) from the System Restore Linux live-cd and it managed to copy off images of both drives without any errors (although it did take a bit of effort to figure out that I had to boot with the restore64 option in order to mount volumes more than 2tb in space with ntfs-3g to be able to mount a volume large enough to dump the images to).  The difficult part was what to do with those images.  My initial attempts of copying the files together and then mounting them failed miserably.  ntfsck ran for 150+ hours before I cancelled it.  chkdsk under Windows also bombed spectacularly on this image (I used Mount Image Pro to mount the dd images under Windows as drives).  I eventually was going to give up and try to just get what I could from the images, and was going to try to do that with Active@ File Recovery.  When I loaded that program I was pleasantly surprised to see a RAID button, that allowed me to select drives and say they were spanned and what order they were initially in.  After that I was able to tell it to copy out all of the old data, and even though that took forever, it did so successfully.  It did have errors copying one trash MP3 I was embarrassed to even own, I don’t know if that was at the junction point or if the file system was somehow broken where it was, but everything else copied out gloriously.

The hardest point of the process was that I couldn’t find any resources of anyone who had successfully done anything like this, so here you go.

Additionally, I freaking love my DroboPro.  I splurged on it because of the angst from my enclosure dying.

edit: Apparently I spoke a bit soon, as Active@ File Recovery doubles as an undelete program, some of the files that had at times existed in more than one place on the drive were messed up when copied off. Booooo. I still don’t know of anything else that would’ve worked. Putting the drives in an enclosure from a different chipset maker certainly is not a solution.