What to do right now
- Power off the NAS, label each disk with its bay number and do not start a repair or rebuild on the NAS.
- Connect the disks to a PC, cancel any offer to initialise them, and image every disk to separate storage.
- Rebuild the RAID virtually from the images, scan it and copy the files to another disk.
Step by step
Stop the NAS and label the disks
Shut the NAS down cleanly if it still responds. Do not run a repair, rebuild or initialisation. Before removing the disks, write the bay number on each one, because the order of the disks is the order of the array.
Connect the disks and cancel initialisation
Attach the disks to desktop SATA ports or to one USB dock per disk. Windows cannot read the NAS file system and Disk Management may ask to initialise the disks; click Cancel every time, because initialising writes a new partition table.
Check health first
Open Disk health & tools and look at each disk's S.M.A.R.T. summary. A disk with reallocated or pending sectors, or one that clicks, should be read as little as possible, so image it first and only once.
Image every disk
Use Create a drive image (skip bad sectors) for each disk and save the images to a separate disk with enough free space for all of them. Unreadable areas are skipped and recorded in a .map file, and an interrupted image can be resumed.
Rebuild the RAID from the images
Open Rebuild a RAID from its disks and add the images in bay order. If Linux RAID (mdadm) metadata is found, apply the suggested level, chunk size and order, check them, then click Build virtual RAID. For RAID 5 with one disk missing, leave that disk out and it is rebuilt from parity.
Scan the virtual RAID
Click Scan this RAID. An ext4 volume opens read-only with its folders and names, so you can copy shared folders directly. Run the Balanced scan as well if you are looking for deleted files.
Copy to a separate disk
Save the recovered files to a disk that is not one of the NAS disks and not the disk holding the images. The app refuses to save onto any member disk of the virtual RAID.
What a NAS actually stores on its disks
Most home and small office NAS units run Linux. Each disk is split into small system partitions and one large data partition, and the data partitions are joined into a software RAID by mdadm. On top of that sits the volume, often ext4 or Btrfs, sometimes inside an LVM layer used for storage pools or flexible RAID types. FileFix Ultimate detects mdadm superblocks and can assemble RAID 0, 1, 5 and JBOD from them.
Flexible schemes such as Synology Hybrid RAID combine several mdadm arrays with LVM. Whether such a volume can be read depends on how it was built; if no file system appears after the virtual RAID is assembled, the data sits in a layer the app cannot read, and you should stop there rather than experiment. Synology and QNAP are trademarks of their respective owners, are named here only as examples, and are not affiliated with FileFix Ultimate.
Btrfs, ZFS, RAID 6 and RAID 10 are not supported
Many recent Synology volumes use Btrfs, and QNAP models running QuTS hero use ZFS. FileFix Ultimate cannot read either file system, nor can it assemble RAID 6 or RAID 10. If your NAS uses one of these, the realistic options are to put all disks, in order, into a NAS of the same type, to use Linux tools made for that file system, or to contact a recovery lab. Imaging the disks first is still worthwhile, because it protects the originals whichever route you take.
Why you should not rebuild the array on the NAS first
A rebuild reads every sector of every remaining disk and writes to the new one. If a second disk is weak, it may fail halfway, and an array that was degraded becomes unreadable. Replacing the wrong disk or accepting an initialisation can overwrite the metadata the virtual rebuild depends on.
Also watch for a disk that dropped out of the array earlier. It holds older data, and mixing it with the others produces corrupted files. If the NAS log shows which disk failed first, leave that disk out of a RAID 5 rebuild.
When it won't work
- The volume is Btrfs or ZFS, the array is RAID 6 or RAID 10, or the shared folders are encrypted.
- Two or more disks of a RAID 5 are missing or unreadable, or one disk of a RAID 0.
- A disk clicks, is not detected or keeps disconnecting; that needs a recovery lab rather than software.
- There is not enough free space on another disk to hold an image of every NAS disk.
How FileFix Ultimate helps
- Free for personal use with no recovery limit, and reads every NAS disk and image read-only.
- Rebuilds RAID 0, 1, 5 and JBOD virtually from disks or images, detects mdadm metadata and reconstructs one missing RAID 5 disk from parity.
- Creates drive images that skip bad sectors, keep a .map file and can be resumed.
- Opens ext2, ext3 and ext4 volumes read-only and refuses to save onto any member disk.
FileFix Ultimate is free for personal use, works offline and never writes to the drive you are recovering from. Business use needs a one-year licence per PC. Check the signature before you run it.
Frequently asked questions
Can I rebuild the RAID from the physical disks without imaging them?
The Rebuild a RAID screen accepts physical disks too, and it only reads them. Imaging is still safer, because each disk is read once and every later attempt works on the images.
Does the order of the disks matter?
Yes. The list order is the array order. When mdadm metadata is found, the suggested order comes from it, but checking it against the bay numbers you wrote down is a good habit.
How much space do I need for the images?
About the full size of every NAS disk added together, not just the used space. Four 4 TB disks need roughly 16 TB of free space on other disks.
Can I recover deleted files from the NAS this way?
Sometimes. After the RAID is assembled, the Balanced scan looks for deleted ext4 files, but ext4 removes much of the information about deleted files, so names often do not come back and signature search becomes the main route. Check the NAS recycle bin folder first if it was enabled.