Taking work now — the first look is freeSpinning disks in a machine Quicker still, give us a ring:0203 0868633
ADR Advanced Data Recovery 0203 0868633 Price my job
ADR / Media and machines / RAID 5 sets and rebuilds that stopped

RAID 5 · a single disk of parity

RAID 5 data recovery, London. Image every member before anyone rebuilds. That decision decides whether the volume comes home.

Parity on a RAID 5 covers exactly one absent disk. What ends the second is usually not a fault at all but the rebuild: one flat-out read from end to end, asked of drives that came out of the same carton and have shared a cabinet temperature ever since. The price first, since almost nobody publishes one: £500 + VAT is where arrays, NAS boxes and servers open, and the exact figure is set down in writing once an engineer has looked at no charge. Your posted disks go home untouched. Disregard the front panel as well — a card reaches the verdict failed in about a second, which is not a diagnosis. Sets arrive from a Harley Street practice with its imaging on a basement NAS, a Park Royal food factory holding batch and traceability records, and a Barnet accountancy practice on a four-bay box in the back office.

On most jobs: no data, no bill A free look first, then one figure in writing Media posted in from Croydon, Ealing and Camden

Put it to an engineer — the first look costs nothing
0203 0868633

RAID 5: what the symptom usually means.

Different fault? Try the finder →
What is going onThe usual reason for itWhere that leaves you
The card reports a member goneThe single disk of parity is spent, and the volume goes with the next faultPower it down before another member goes
A rebuild ran half the night and then stoppedA disk of the same age met a soft patch during that full read — much the commonest endingLeave it switched off. Do not go again
Two members marked failedAlmost always one finished disk next to one merely slow across a few sectors, not a matched pairSend the whole set; the images separate them
A dropped disk was pushed back inStale blocks are now mixed in with live stripesDo nothing else to it
A replacement controller sees no arrayThat is a disagreement over metadata. Not one of your files has shiftedReassembly, and nothing worse
Recovery software has been aimed at the membersWrites have gone onto the disks you still holdSend them regardless, and name the tool that ran
Packing it and posting it:box it so nothing can move about, and put the parcel's insured value at what those files are worth to you, not what the hardware cost. Send it tracked; parcels go to the intake lab, and the leg home is paid at this end. Want a word first about what goes in the box? Ring us before it is sealed and an engineer will run through it. You will find the lot spelled out on theguide to packing and posting.

What a rebuild asks of the disks left in the chassis.

How much a rebuild has to readA rebuild has one speed and it is flat out. Every disk left in the chassis gets read right through: four 16TB members means the three still standing owe roughly 48TB between them, each carrying a running-hours count that matches the disk which stopped.
How to read the URE figureThe consumer specification is one unreadable sector per 1014 bits, about one awkward sector per 12.5TB. Enterprise kit is specified at 1015, ten times the headroom. Read it as the least a manufacturer is willing to stand behind, never as a countdown, and then notice how many drives in service run well past it. Nothing ails the number. What has been built on top of it in graph form is a separate matter.
What the card does about itIt depends on the card's age. An older controller reaching a sector it cannot finish discards every hour of rebuild already done. Newer firmware, on PERC and MegaRAID, marks that stripe bad and keeps going; mdadm under Linux notes the bad block and behaves the same way. Giving up one stripe for one unreadable sector is worth it. Giving up a whole volume never was.
Why the next one followsOne carton, one chassis, one shared history of temperature. Ask for a single vast uninterrupted read of every disk and whichever was quietly failing fails there and then. The same ground is covered slowly here, with the soft patches deliberately left until last.

From the box arriving to your files going back.

Work we have closed →
01

Logged the day it lands, and the first look costs nothing Free

A case number goes on it the day the parcel lands, and an engineer settles what has truly failed before anything else happens — free of charge, and first in the order of work. Back to you come two things together: a straight note of what is liftable and what is not, plus one figure, fixed and written down. Accept it, or decline and owe us nothing.

Nothing to pay for lookingA single figure, put in writingNothing owed at this stage
02

Take an image of every member

Every member is copied whole onto storage of ours, including any the card had rejected. Any read that comes back slowly gets noted while it happens, and the roughest areas are saved for the end. Your posted disks receive no writes whatever.

An image of each member firstMembers the controller had dropped included
03

The layout read off the disks

Three things get settled before assembly: the bay each member occupied, how big a stripe is, and the rotation direction of parity. Every answer comes out of what the disks themselves recorded, and is then tested against how your data actually sits. Guesswork plays no part.

Bay order and stripe size establishedSettled before anything goes together
04

Build the array over the images

The array goes together on top of the images, never on your disks. Each block comes from whichever image surrendered it most cleanly, and where none did, parity supplies it — which is how material sitting on a member the card had rejected still reaches you. Repairs to the file system follow, then the volume is mounted and read through before anything is released.

Every block from its cleanest imageThe volume mounted and checked
05

You see the file list before you pay

What was recovered is listed for you first, and only then does a bill exist. Approve the list and it is invoiced; turn it down and it is not — and where nothing has come back, most jobs carry no charge whatever. Recovered data travels home on fresh media bought in for your job, with the postage at our end. Your case is not closed until you have read those files on a machine of your own.

No charge until you accept the figureFresh media, supplied with the jobThe post home is ours

What arrives most often

  • Degraded is a description of the past — one disk of cover existed, and the member that left carried it off. From that point the array is as exposed as a lone drive with nothing standing behind it.
  • Array recovery is sold on fear — this is the dearest work on any bench, and much of what gets written about it is designed to unsettle you. What you get here is whatever the imager finds on your own disks, written down.
  • A puncture costs a single stripe — reaching a sector it cannot read, present-day firmware marks that stripe punctured, logs it and continues; an older card would discard the whole rebuild at that moment. Once the array is assembled, every punctured stripe can be listed for you by address.
  • Linux software arrays keep a record — mdadm superblocks read alongside the bad-block log will show which disk left the array, and roughly when. Hardware cards are a great deal less forthcoming.

The arithmetic, once: the consumer rating is one read error the drive cannot correct in 1014 bits — roughly one sector every 12.5TB — where an enterprise disk is rated ten times better again, at 1015. Four 16TB drives means a rebuild needs some 48TB read faultlessly off the three that remain, so this is a sum rather than a sales tactic. Against it, drives in service routinely outlast what they are rated for. Neither statement cancels the other, and both argue for the same running order: images, then the rebuild — since after imaging a rebuild costs nothing.

One job, followed all the way through.

LDN · ADR-2026-3372JOB LOGGED ✓

Two RAID 5 members down, and the year-end still filed on time

Nineteen bad sectors, and that was all of it. The set looked as though two disks had gone together; in truth one member had dropped out months before without anyone noticing, so once a second stopped partway into the rebuild nothing was left holding the volume up. Copying the pair showed how little of that second disk would not read, and the stripe was worked out from three sound members and a fourth that was very nearly sound. The year-end filing still went in on time.

19unreadable sectors, and no second failed disk6days here, and back to you

What helps, and what harms.

Do this much first

  • Cut power to the chassis as soon as a member drops
  • Mark each disk with its bay position before you pull it
  • Send the entire set, rejected members included
  • Tell us the controller, and everything already tried

What sets us back

  • Kicking off a rebuild with a member absent
  • Shoving a rejected disk back online out of the card's own menu
  • Turning off-the-shelf software loose on the disks
  • Slotting in a spare and letting the card initialise

Questions answered before you commit.

The card shows two members failed. Have we lost the volume?

Rarely. Controllers drop members on their own terms — a few sectors answering marginally late will do it — and being dropped is not at all the same as being finished. Nothing is attempted until every member has been imaged onto storage of ours; parity from the rest then supplies whatever one image will not yield. Two members marked bad by a card is an ordinary week here. Two genuinely past imaging is unusual.

We are degraded and one member short. Should we rebuild?

Not before images exist. A rebuild orders every disk left in the chassis to read itself end to end, which is precisely what a tired drive cannot do, and it writes new parity over the old — and the old parity was what would have reassembled your volume. Take images first and rebuilding afterwards costs you nothing, because no later operation reaches an image.

Are the URE numbers worth worrying about?

A URE is one sector the drive's own error correction cannot repair, at which point it halts and reports rather than return something false. The consumer specification works out at roughly one such sector per 12.5TB read; enterprise is one per 125TB. Treat those as the floor a manufacturer will put its name to, not a countdown. Modern firmware marks the stripe bad and carries on.

Do the disks need the controller with them?

No, the disks on their own are enough. Every member records the way the set was built, and where that record is damaged the shape of your data answers the same question. Assembly is done in software over the images, nothing is written to your originals, and the card stays where it is bolted.

Begin here if yours is doing the same thing.

RAID and server jobs → Servers that refuse to come up → Nothing detects the disk → Every device we take →

Nothing gets worse while the power is off.

Looking at it is free. Back comes a list of what opened and what did not, together with a single price to finish, set down in writing while you are still free to say no. On most jobs an invoice only follows the data. Until that list reaches you, leave the drive switched off.

0203 0868633

Get it assessed Sending it to Guildford