Data recovery for RAID and servers
Intervention on complex RAID systems, enterprise servers and critical configurations with data loss or multiple failures.
Request diagnostic →Advanced RAID array reconstruction
RAID systems require complete logical reconstruction: disk order, parity, blocks and internal structure.
One mistake in the process can permanently destroy the data.
RAID types
- • RAID 0, 1, 5, 6
- • RAID 10
- • NAS and servers
- • Hybrid configurations
- • Virtualised systems
Common scenarios
Recovery process
System diagnostic
RAID identification
Logical reconstruction
Safe imaging
Data extraction
Related services
Symptoms of an array you should no longer touch
Modern RAIDs are designed to survive one or two disk failures. When the symptoms below appear, what was redundancy becomes a risk: any in-place rebuild attempt can destroy the rest of the array.
Two or more disks down
RAID 5 with two failures, RAID 6 with three. The array is offline. Do not start rebuild under any circumstances.
Rebuild stalled or failed
Reconstruction stuck at 47% for days. Usually indicates unreadable sectors on another supposedly healthy disk.
Dead controller
PERC, Smart Array or LSI that doesn't power on or doesn't detect the disks. The array is intact but reading is impossible without virtual reconstruction.
Lost configuration
The NAS shows disks as 'uninitialised', the system asks to format, or the array appears as JBOD. The metadata is corrupted.
Stripe, parity, metadata and why we never touch the original
A RAID array is not the sum of its disks: it's a logical map that defines how blocks are distributed among them. That map includes the disk order, stripe size (typically 64 KB, 128 KB, 256 KB), parity rotation (left-symmetric, right-asymmetric, etc.) and the position of metadata on each disk (at the start, at the end, on sector 0, on an LBA reserved by the controller). Without all these exact parameters, the data is there but unreadable — like having all the pages of a book shuffled and unnumbered.
The parity is what allows a RAID 5 to reconstruct the content of a lost disk from the XOR of the rest, and a RAID 6 to survive losing two. But that same parity becomes useless if the surviving disks have marginal sectors: when the rebuild tries to read every block to calculate the new disk's content, a single URE (unrecoverable read error) on another disk can abort the reconstruction and leave the array doubly degraded. It's the classic failure of RAID 5 with large disks.
That's why our first operation is always the same: individual cloning of each disk onto healthy targets, in forced-read mode with error tolerance (ddrescue, deepspar). From the cloned images we reconstruct the array virtually — trying combinations of order, stripe and rotation until the filesystem validates — and extract the data. The original disks are never connected to the original controller, nor allowed to attempt any rebuild.
Mistakes that destroy recoverable RAIDs
Almost every array that arrives at the lab in an irrecoverable state does so after a well-intentioned previous attempt. These are the most expensive — and the most common.
Launching a rebuild on a degraded array
If another disk has marginal sectors, the rebuild finishes it off. Clone first, rebuild later.
Changing the disk order
Hardware controllers identify disks by bay number. Mixing the order makes the array unrecognisable or, worse, initialises it incorrectly.
Accepting the NAS 'initialise disks' prompt
One way or another, the NAS will lose the array configuration and the data. If it asks to initialise, power off and isolate the disks.
Running fsck or chkdsk on the array
These utilities rewrite filesystem metadata. On a poorly reconstructed array, they can permanently destroy the directory structure.
Replacing the controller with a different one
Each manufacturer has its own metadata format. Connecting PERC disks to LSI or vice-versa can overwrite the headers.
Forcing the array online ignoring warnings
When the controller marks the array as inconsistent, there's a reason. Forcing online without prior diagnostic is the recipe to lose everything.
RAID levels and systems we recover
We work with hardware and software arrays, classic enterprise configurations and home NAS. If your setup isn't listed, write to us: we likely cover it.
RAID 0 / 1 / 10
Stripes, mirrors and combinations. No parity but with full dependency on block order. Virtual reconstruction from cloned images.
RAID 5 / 6
Rotating parity (single and dual). Automatic identification of order, stripe and rotation pattern. Recovery with two missing disks on RAID 6.
RAID 50 / 60
Nested configurations common in enterprise servers. Two-level reconstruction: inner arrays first, outer stripe afterwards.
Synology SHR / SHR-2
Synology proprietary RAID based on mdadm + LVM + Btrfs/ext4. Full support including expansions with disks of different sizes.
QNAP, WD My Cloud, Buffalo
NAS with mdadm+ext4, ZFS or proprietary RAID. Btrfs snapshot export and degraded ZFS pool reconstruction.
Hardware: PERC, Smart Array, LSI
Dell PERC H7xx/H8xx, HP Smart Array P4xx/P8xx, LSI/Broadcom MegaRAID, Adaptec, Areca. Metadata reading without the original controller.
What we get asked the most about RAID and servers
What if two disks fail simultaneously in a RAID 5?
+
Why can a RAID rebuild make things worse?
+
Do you recover Synology, QNAP, WD My Cloud or Buffalo NAS?
+
What information do you need to start the diagnostic?
+
Can I keep using the disks while waiting for the shipment?
+
How long does it take to recover a RAID with 4 or 6 disks?
+
Do you work with hardware RAID (Dell PERC, HP Smart Array, LSI)?
+
Do you sign NDAs for companies?
+
Recover your RAID system risk-free
Specialised intervention in critical environments. Free evaluation.
Request diagnostic