We can recover broken RAIDs from the PROAVIO Enhance desktop RAID system. We have just completed a RAID 5 recovery where the unit was plagued by disk failures. Despite previous disk failures being handled successfully, during the replacement & rebuild of one failed Seagate ST2000DM001 disk, another disk failed at the same time. This left all 7.5TB of the data unreadable.
These Seagate disks are becoming notorious for failures, so our first step was to make full clones of every disk. Even the working ones. This protects us from any further disk failures during our recovery process. We use the copies from that point onwards and put the original disks safely away.
After checking the data on each disk, we determined that the top disk was only partially written, so was likely in the middle of a rebuild when the other disk failed. We had to discount this disk as it didn’t contain enough of the RAID to rebuild the data.
The next disk down had a serious fault (a failed head), and was only partially accessible. After carefully imaging as much of the disk as possible, we had to take this drive apart in our cleanroom & replace the heads. This allowed us to gain access to the previously missed areas of the disk to get a full image.
The rule when rebuilding RAID 5 is that you can only have a single disk failure at any one time. In our case, the disk was incomplete and had to be ignored, so we were left needing the second failed disk. Fortunately the failure could be overcome so we then had four of the five disks available for a RAID rebuild.
Analysis of the individual disks allowed us to determine which stripe size, parity alignment, and disk order had been used. This information varies by vendor and is often either secret, or unknown. Adding this information to a virtual RAID Controller didn’t allow immediate access to the files. We next had to scan for the HFS+ filesystem structures, which had been damaged during the RAID failure. Once the Mac filesystem was rebuilt we were able to start extracting files. Our virtual RAID Controller filled in for the missing disk as required using the parity information from the other disks.
In the end, our experience with the ST2000DM001 disks, our knowledge of Mac filesystem structures, and our RAID Recovery skills allowed us to extract the crucial data within days of receiving the damaged system.