Recovering a Drobo 5N
I have had the disks from my Drobo 5N sitting in protective cases for five years. We had a power cut while it was doing something and from that point I was unable to connect to it despite everything I tried. I did get in contact with a specialist Drobo recovery business who quoted over £1k to restore the files, plus I would need to need to ship it to them and would probably need some other storage solution. So I delayed. Meanwhile the company shut, I found another closer so we could drop off the disks.
Fortunately most of the important files needed were on the cloud and so over time I forgot about the disks. This week I questioned Claude if I could recover the files myself and the answers it gave me were strong enough for me to purchase a new Segate 24Tb drive (how big???) and a cheap hard drive dock.
Yesterday I was able to make disk images of all five 2TB disks I had used in the Drobo and delightfully they had no bad sectors, no errors and all the drives span up without issue - left too long and they can seize up.
Why a defunct Drobo is a particular problem
Drobo used a proprietary storage layer called BeyondRAID. It was not standard RAID and it was not documented publicly. I really liked the plug and play and ability to swap out bad drives solution that the Drobo provided - but this one died on me with no simple way of recovery so I'm not going back - and of course the company went bankrupt in 2023.
What not to do
Before touching anything I had a list of things that could make the situation worse that I needed to be aware of:
- Do not factory reset or "repair" through Drobo Dashboard. With no manufacturer infrastructure left, this is close to unrecoverable.
- Do not add a replacement drive. That triggers a relayout, which rewrites the array.
- Do not let anything write to the member disks. Including well-meaning operating system prompts.
- Do not click Initialise. macOS will offer, repeatedly. See below.
- If a drive will not spin up, or sounds wrong, stop. That is the one failure mode software cannot fix, and every retry makes professional recovery harder.
Not a problem in my case as I had removed, and numbered, all the drives ready to send off somewhere to be recovered.
The approach
The plan was to avoid touching the original disks any more than absolutely necessary, and to do all the actual recovery work against copies.
- Buy a destination drive large enough for full disk images plus the recovered data. In my case I opted for 24Tb
- Buy a plain SATA dock with no RAID or clone functionality.
- Read SMART from each drive, then image it, in a single power-on session
- Reassemble the array from the images, not the physical disks
- Extract to a normal filesystem
Five 2TB drives meant 10TB of images. Usable capacity on a five-disk BeyondRAID with single-disk redundancy is around 8TB. A 24TB desktop drive covered both with room to spare, and became an archive drive afterwards.
For the dock, the deliberate choice was one without offline clone. A one-button disk-to-disk copy function sitting next to five irreplaceable drives is a hazard, not a feature. Pure JBOD passthrough, UASP, its own power supply, around £23. I got a double port one as it was on offer and cheaper than a single port one! I only used a single port.
Imaging
For the 10Tb of data I had this took about 24 hours.
The single most important decision: one spin-up per drive. Drives that have sat unpowered for years are at their most vulnerable during spin-up, not during reading. So SMART and imaging happen in the same session, then the drive gets parked. I was lucky - they all worked.
The numbers that mattered were reallocated sector count, current pending sector count, offline uncorrectable, and CRC errors. All five of mine came back at zero, with around 37,000 power-on hours each. For five ten-year-old NAS drives, that was a clean sheet.
Danger point!
Each time I sat a drive in the dock and powered it on the Mac OSX would pop an alert up telling me that the disk was unreadable, which being a Drobo disk Mac OSX cannot read it of course, and asked me if I wanted to:
- Eject
- Initialise
- Ignore
Ejecting doesn't help, Initialising will reformat the drive loosing everything!!!! Ignore is what I wanted. Its obvious really but alarming the first time you see it.
Reassembling the array
With five images in hand, the tool I chose was UFS Explorer RAID Recovery, which supports BeyondRAID and runs natively on macOS. The trial has no time limit and will scan and display the full file tree so I could check that I could actually recover files before purchasing a personal licence. It is tricky software to use as a Mac user but with a bit of patience is useable.
First I had to assemble the five drive images as disk images so it was treated as a single disk. Once all five were attached, each showed a "Drobo BeyondRAID partition" beneath it — the first real confirmation that the format was recognised on each part.
The meu offered a Drobo BeyondRAID assistant which I then ran. Twenty minutes of reading array metadata produced a single configuration: 5793 GB allocated, single redundancy, five of five disks present.
The result
Using Explore on this new drive the file tree opened directly, no scan was needed. Named folders, correct dates, intact hierarchy, and image thumbnails rendering in the preview pane — which proved file content was readable, not just directory entries.
Nothing missing - an absolute win!
I then chose a directory to save, basically all the files we had put on there, which weighed in at about 2Tb and that took 23 hours to save over 2.5million files to the new disk. While running the save I did get an alert telling me that it was about to save a file, of zero bytes, but there was already a file of the same name with an amount of bytes and gave me a choice. I chose to save both files and ticked the box to do the same with any others, which on a long run like this is important. I can find and delete the zero byte ones later. in the end only 1 file failed to process.
There are a set of Drobo Apps I had installed but that folder has no data I want to recover and there are two other folders which were our Mac TimeMachine Backups. I am considering saving mine as I remember deleting some images that I wish I had not and they may be in there. That directory is 1.5Tb though so I might see if I can save parts of it, else it will take another day. Once done I can delete the images from the new drive freeing up 10Tb of space. I still have the drives which I might do something with - perhaps another Raid device.