A RAID array that has dropped offline is usually recoverable — the data is still on the disks. What decides the outcome is what happens next: the single biggest cause of permanent RAID loss is a DIY rebuild onto a still-degraded array. Here’s what actually fails at each level, the don’ts that matter most, and how a failed RAID 0, 5 or 6 is rebuilt safely.
The data is still there. A degraded array can almost always be reconstructed — but only if it isn’t written to first. A rebuild, a re-initialise or a “repair” is what turns recoverable into lost.
The array stops assembling — but your data is still sitting on the platters.
A RAID spreads your data across several disks in fixed-size blocks, often with parity — extra maths that lets the array survive a disk failure. How many disks it can lose depends on the level: RAID 0 can lose none, RAID 5 can lose one, RAID 6 can lose two. Lose more than that — a second disk in a RAID 5, a dead controller, a corrupt configuration, or enough bad sectors across several drives — and the array can no longer assemble itself and drops offline.
The crucial point: offline doesn’t mean gone. The blocks are still on the individual disks exactly where they were written. Recovery is the job of reading each disk, working out how the blocks were arranged, and reassembling them — away from the original hardware, so nothing is ever written back to your data.
Each RAID type fails and recovers a little differently.
RAID 0 (striping). No redundancy at all — data is split across every disk for speed, so a single failed disk takes the whole array down. Recovery needs every member: we recover any physically failed drive first, image them all read-only, then rebuild the stripe from the images by working out the block size and disk order.
RAID 1 (mirroring). Every disk holds a full copy. A single disk failure is survived outright, and even if the array won’t mount, we can recover from either healthy member. Trouble usually means both disks have issues, or the file system on the mirror is corrupt.
RAID 5 (striping with parity). Survives one failed disk by rebuilding it from parity. The classic disaster is a second disk dropping — often during a rebuild — which takes it offline. It’s still usually recoverable: we image every disk, including the failed ones as far as they read, and reconstruct the array using the parity and the surviving data.
RAID 6 (dual parity). Survives two failed disks, so it takes three failures to go offline — but the same rules apply, and the same rebuild dangers. We treat it the same way: image everything read-only, then reconstruct.
RAID 10 (mirrored stripes). A stripe of mirrors: it tolerates a disk failure in each mirrored pair, but losing both halves of one pair loses the array. Recovery combines the RAID 1 and RAID 0 approaches — recover the failed members, then rebuild the mirrors and the stripe.
Almost every unrecoverable RAID we see was made worse after it failed, not by the original fault.
Don’t rebuild onto a degraded array. If a disk has dropped, replacing it and letting the controller rebuild stresses every remaining disk and writes new parity across all of them. If another disk is on the edge, the rebuild finishes it off — and takes your data with it.
Don’t re-initialise, recreate or “repair” the array. Recreating the array or accepting a controller’s offer to initialise it can overwrite the configuration and the data layout. Once that’s gone, reconstruction is far harder.
Don’t run chkdsk or fsck on the array, and don’t keep forcing it back online. Repair tools write to the volume and can turn a clean rebuild into a partial one.
Don’t swap disks around guessing the order — and before you remove anything, label each disk with its bay position. The physical order matters, and a mislabelled set costs time to untangle.
Read everything, work out the geometry, reassemble a copy — never the originals.
The professional process is the opposite of a rebuild. First, every member disk is imaged read-only, sector by sector, on write-blocked hardware — any drive that is physically failing is repaired or head-mapped first and imaged around its weak areas, so nothing degrades further. From that point the original disks are set aside; all the work happens on the images.
Next we reverse-engineer the array’s geometry from the data itself: the stripe (block) size, the order of the disks, the parity rotation and whether it’s left- or right-synchronous, and the start offset. This is why a dead controller or a lost configuration isn’t a dead end — the layout is written through the data, and it can be worked out from the images.
With the parameters known, we rebuild the array virtually from the images and recover what sits on top — the file system and its folders, or the virtual machines (VMDK, VHDX) and databases the array was hosting. Our RAID 10 explainer and the guide on what to do when a RAID won’t rebuild go deeper on the specifics.
A failed rebuild, a second dropped disk, or a business-critical array down — power off and let us image it.
If a rebuild has already failed, a second disk has dropped, the controller no longer sees the array, or the system simply won’t come back and the data matters, the safest move is to power the system down and stop trying to bring it online. Every extra attempt risks writing over what’s still recoverable.
Our RAID data recovery service handles arrays and NAS units in-house from a fixed £500 +VAT, with a free diagnostic and a written quote before any chargeable work, and your disks never leave the UK. Remove the disks, label each with its bay order, pack them so they can’t knock together in transit, and send them in — business-critical arrays are prioritised.
Usually, yes. RAID 5 survives one failed disk; a second failure takes it offline, but the data is still on the platters. We image every member disk read-only — including partially failed ones — work out the stripe size, disk order and parity rotation, and rebuild the array offline from those images. How complete the result is depends on how much of each disk still reads, which the free diagnostic tells us.
Stop and power the system down. A rebuild forces every disk to be read in full and writes fresh parity across the array; if a second disk is already weak, that can overwrite the very data needed to reconstruct it — the commonest cause of permanent RAID loss. Don’t retry the rebuild or re-initialise the array. Label each disk with its bay order, remove them, and send them in; we image every member read-only and rebuild the array offline.
RAID 0 has no redundancy, so every disk is needed — the data is striped across all of them. If one disk has physically failed we recover that drive first (heads, board or firmware work as needed), image every member read-only, then rebuild the stripe from the images by working out the block size and disk order. As long as the failed disk can be read, the array can usually be reassembled.
Usually not. We reconstruct the RAID parameters — stripe size, disk order, parity rotation and start offset — directly from the data on the imaged disks, so a dead controller or a lost configuration doesn’t stop recovery. The original controller or a note of the setup can help, but it isn’t required.
RAID, NAS and server recovery starts from £500 +VAT. The exact figure depends on the number of disks, the RAID level and the capacity. You get a free diagnostic and a fixed written quote before any chargeable work, and everything is carried out in-house.
It varies with the number of disks and their condition — imaging several drives read-only, especially any that are failing, takes time. The free diagnostic usually completes within 48 hours, and business-critical arrays are prioritised, with an honest turnaround given once we’ve seen the disks.
Power the system down, tell us the make, the RAID level and what happened, and send the disks in — you’ll get a free diagnostic and a fixed written quote before any money changes hands, with business-critical arrays prioritised.