Home / RAID 1

RAID 1 Mirror Recovery

Two disks that reached the end of their working lives within a fortnight of each other. A rebuild that ran backwards and copied the stale member over the live one. A box that died with two healthy disks locked inside it. None of those is a lost cause. Both halves are imaged separately, weighed against each other on what they actually contain rather than on a status flag, and your files are lifted from whichever turns out to be the better source.

No part of a mirror job is guessed at. The examination costs nothing and the written figure comes out of it — from £500 + VAT on disk count, logical work under no fix, no fee.

// how mirrors arrive here

Two disks present, and nothing to open

A mirror gives one undertaking and keeps it well. Every fault below goes around that undertaking rather than through it, which is how pairs of perfectly ordinary disks end up in a padded box heading south on the M40.

The enclosure died and both disks are fine

In a two-bay unit the first thing to fail is usually the bridge chip, the controller or the power supply rather than either disk. That leaves two serviceable members shut inside a case with nothing left to talk to, and the only way forward is to read each of them directly, outside the box.

A pair bought together and worn out together

Two disks from one order, switched on the same afternoon and given identical work ever since, rarely fail years apart. The gap is more often weeks. Whatever wore the first one out has been doing the same to its twin throughout, at the same temperature, in the same cupboard.

A rebuild that ran in the wrong direction

Put the wrong disk back in a bay, or leave the unit to reach its own conclusion overnight, and the stale member gets copied over the current one. The hardware performs faultlessly. It simply performs backwards, at full speed, without pausing to ask.

The disks spin and the share has gone

Both members healthy, both members readable, and no volume presenting itself to anything. The damage is above the mirror — the volume manager or the file system laid on top of it — and the disks underneath are carrying the data perfectly well.

Two halves holding two different editions

Where a pair has run degraded at separate times, each disk ends up with its own version of the same folders. Merged carelessly they spoil one another. Copied apart, each is free to give up the material only it still holds.

The mistake arrived on both disks at once

A deletion, an overwritten save, an update that went wrong, an encryption run. Mirroring exists to reproduce writes the instant they happen, so all four reach the second member at the same moment as the first and neither disk is spared any of it.

// the one promise, and its edges

The single sentence a mirror is able to guarantee

The arrangement is simple enough to describe in a line. Two disks are kept identical down to the sector; each write goes to both; and the operating system is not told the write finished until both have accepted it. What deserves attention is the part the design has no opinion on. At no stage does anything ask whether the write was a sensible one.

Which is why the list of disasters a mirror prevents is one item long: a member failing while its partner carries on serving. Emptying a folder is a write. So is a bad save, a database closing down badly, an update that misfires, an encryption run working through a share. Every one of them is duplicated onto the second disk at bus speed with no window in which anybody could intervene. Beyond writes, the shared parts finish the argument — one power supply, one controller, one cupboard. A fire takes both. A theft takes both. A dead supply stops both at the same second.

Held against a backup the distinction is easy to state. A backup is a copy of an earlier moment kept somewhere else. A mirror is a copy of this moment kept in the same box. Both are worth having and only one of them can return you to yesterday. And to be plain about what is being offered here: this site sells no backup product, nothing to sign and nothing billed monthly. It is a bench that gets data back afterwards, so anything it says about where copies belong costs you nothing.

Why the box's own vote is not trusted at this end

Every member keeps a short housekeeping record of its own: an event counter that advances as things happen to the set, a timestamp, and a flag saying whether that disk is currently in step with its partner. When a set assembles, the two records are compared and the disk showing the higher counter is treated as authoritative. The other member is then brought into line with it, and on a mirror being brought into line means being overwritten.

Most of the time that logic is correct. A disk that spent a fortnight out of its bay genuinely is behind its partner, and copying the live member across is precisely what should happen. It goes wrong when the housekeeping records are missing, damaged or misread — a state that turns up regularly after a controller has died or an update has failed part way. The comparison still runs, still reaches a conclusion and still acts on it at full speed. The only difference is that the conclusion is now false and the disk being treated as authoritative is the out-of-date one.

None of that arithmetic is repeated at this end. The two members are imaged first, and only then is a leading copy chosen, on evidence: which image holds a file system that stands up to inspection, which holds the edition of a document with a date you recognise, which shows the tell-tale progression of having been written over from the start of the disk forwards. A counter takes a second to read and can be wrong. Reading two disks properly takes hours and is much harder to mislead.

Where the mirrored pairs in this county live

Most are two-bay boxes — Synology, QNAP, Netgear, Buffalo, WD — with a scattering of small towers under desks. They come from accountants and surveyors in Warwick and Kenilworth, from design and games studios around Leamington, from hotels and letting agents at Stratford-upon-Avon, from engineering suppliers up towards Nuneaton and Bedworth, and from logistics offices on the parks at Ansty and Ryton. Both disks travel in one parcel to Oxford Data Recovery, John Eccles House, Oxford Science Park, Robert Robinson Avenue, Littlemore, Oxford OX4 4GP, tracked and insured. Parity sets are covered on RAID 5 data recovery, striping on RAID 0 data recovery, the wider view on RAID recovery and NAS recovery, the towns on data recovery Warwickshire, and the figures on data recovery cost.

// what a healthy-looking mirror hides

The member that goes bad without mentioning it

Either half can satisfy a read, which sounds like a benefit and hides something slow. Sectors do not fail all at once; they weaken. A unit with no scrubbing routine has no reason to touch a quiet corner of a disk for months on end, so one member can be gently rotting behind a front panel showing two green lights. The symptoms, when they eventually appear, are odd rather than alarming: a file that opens on Monday and will not on Thursday, a checksum that passes twice and fails the third time, an application complaining about data it produced itself. If that is the story that brought you here, both disks matter more than usual, because for any given file the intact copy may survive on only one of them.

One parcel, two separate jobs

The pair is separated as soon as the box is opened and worked as two independent recoveries. Each member goes onto a hardware imager on its own, and any disk carrying a physical fault is stabilised in the clean-air enclosure before that starts. Only afterwards are the two images set against one another and compared for content: which of the two carries more of your material, and which carries the more recent version of it. One becomes the base for the rebuild and the other supplies whatever the base is missing. At no point is anything written to either original.

Hardly any two-bay box runs a textbook mirror

Synology surrounds the standard Linux arrangement with a volume manager of its own and puts a modern file system over the result. QNAP does something similar under different names, Netgear's X-RAID carries expansion logic of its own, and Buffalo has variations again. Every one of them can be read with the right tooling. None of them can be read by pushing a member into a Windows machine and hoping. What that asks of you is very little: send both disks rather than the one that looks guilty, note which bay each came out of, and say what make and model the unit was.

The ten minutes after you notice

A rebuild heading in the wrong direction costs a fraction more with every write it completes, which makes the interval between noticing and pulling the plug the most valuable ten minutes described anywhere on this page. Cut the power, and cut it now rather than after one more look at the web interface. Then leave the disks be. Testing each member in a desktop machine to see which one works buys you almost nothing: at best it confirms what the images were going to show anyway, and at worst a dialogue appears asking whether the disk should be initialised and somebody in a hurry agrees to it. Three habits are worth more than any of that — photograph the front of the unit before a disk moves, write the bay number on each one as it comes out, and refuse every offer the box makes to repair itself.

// related pages

Nearby on this site

// what mirror owners ring up about

Common questions

Seldom. For two members to fail in the same way at the same instant is rare enough to be worth commenting on when it happens, so once both have been imaged — after clean-bench work where a disk needs it — one of the pair nearly always proves the better source. The volume is then raised from that image, with the other filling in whatever it is short of.
The disks alone. An enclosure failing rarely does any harm to what was sealed inside it, so each member is read on its own and the volume is put back together from those images afterwards. Only a genuinely proprietary format makes the case worth the postage, and one call to 0800 689 0668 settles that question before you go hunting for a box.
Frequently quite a lot. A member that has been overwritten usually keeps recoverable material underneath, and its partner still holds whatever it was holding. What the outcome turns on is how far the rebuild got, since every write it completes takes a little more away — which is the whole argument for cutting the power the instant you notice.
From £500 + VAT, tracking the number of disks that have to be copied. Where both members are sound and the fault is logical, no fix, no fee applies as normal. A disk that has failed electrically or mechanically needs donor parts before a single sector can be read, so half the quote falls due at that stage and the balance only once the data is proved back.
It is not, for the reasons above. The one event it answers is a single disk failing. Deletions, corruption, a botched update and an encryption run all reach both members together, and a dead controller or a stolen box removes the pair at once. Protection means copies on separate media kept elsewhere, and this bench does not sell those. It recovers data; it does not store it.
A tracked parcel out of Warwickshire generally reaches the bench the next working day, and the free assessment closes 2 working days after the disks are booked in. From your approval, a single-drive recovery runs 2 to 4 working days and an uncomplicated mirror is usually done inside the week. Anything needing the clean bench, and anything very large, runs beyond that, and the written quote says as much when it is issued.
It works now and then and it is still a poor bet. Almost every two-bay unit lays its own volume management over the mirror, so a desktop machine finds a disk it has no way of reading and proposes to initialise it — and one accepted proposal does real damage. Even when a member does mount, you have picked a source with no way of knowing whether it is the current half or the stale one. Send the pair and let the evidence decide.

Mirror gone quiet? Send both disks.

Each half copied on its own, the better copy chosen on evidence, the other one used to fill the gaps. The examination costs nothing.

// getting your media to Oxford

Posting a device in — what goes in the box

Very little of this work changes hands in person. It comes by post, which treats hardware that has already stopped behaving far better than a car journey does: padded, tracked and insured for the whole trip. Coventry to the Oxford bench is about fifty-five miles down the M40, roughly an hour, and a parcel handed over in Warwickshire this afternoon is normally booked in the next morning.

As a rule the storage comes out and the machine stays where it is. That applies to a laptop, a tower, an iMac and to the recorder sitting under a counter. Taking equipment apart is not something this bench does, and a repair shop will free a drive in a few minutes. Two things go the other way: an external drive stays sealed inside its own case, and a NAS travels as a complete unit with its disks still in their bays. A Fusion Mac is a third case — both of its drives come out and travel together, each one labelled. The single situation nobody can work around is memory soldered flat onto a mainboard, which is how Apple Silicon Macs and a good many slim laptops are built: if the storage will not unbolt, there is no parcel to send.

  • Use a box or padded mailer with some rigidity to it, and pack around the drive until nothing shifts when the parcel is tilted. Mains adaptors, docks and leads are not wanted at this end.
  • Sending a RAID or a server? Only the member disks travel — not the chassis, not the controller — and each one wants its bay number written on it. Take a photograph of the front of the unit before anything is pulled; it costs nothing and now and again it saves a day.
  • Print the shipping and booking-in form (PDF), add a name, a number you will answer and a sentence on how the trouble began, and drop it in beside the media.
  • Most people use Special Delivery, which is tracked and covered; a courier of your own does the same job. You can also bring it: the Oxford reception takes devices over the counter, Mon–Fri 9:00am–5:30pm. Neither a Coventry counter nor a collection round exists.
// write this on the label

Oxford Data Recovery

John Eccles House
Oxford Science Park
Robert Robinson Avenue
Littlemore, Oxford, OX4 4GP

↓ Print the shipping & booking-in form (PDF)

Put Oxford Data Recovery on the label. From Coventry it is roughly fifty-five miles straight down the M40, about an hour if you would rather drive it in than post it. Either way you are told the moment it is logged, and the free diagnostic finishes two working days later.

Unsure what ought to go in the box? Ring 0800 689 0668 before you seal it, or work through the free online diagnostic and let it do the asking.