RDR / What we take in / RAID 1 mirrors

Specialist · RAID 1 mirrors

RAID 1 data recovery, Reading. A mirror fails quietly: both sides can be wrong for months before it says so.

The cost of recovering a RAID or NAS begins at £500 + VAT, and the first diagnosis before this is free. With a mirrored configuration data is written twice, so normally both drives in the pair will hold the complete volume. That is why the failures we generally see are very different from what most users expect. Disks purchased at the same time tend to age and fail at similar times. An older controller may write the identical corruption to each side of the mirror. A rebuild can simply copy the failing member onto the remaining good disk. And a mirror which lost a member in the spring has been operating without any form of redundancy since then. Both members are cloned read-only individually, and then the clones are compared block by block. The volume is then created using whichever member reads successfully and whichever holds the latest version. The disks arrive at our Oxford intake lab by post from Reading, Theale and surrounding areas such as Woodley, Earley, Newbury, Didcot and Basingstoke. Your original disks are then returned to you unchanged.

Only pay if we recover your data, on most jobs The diagnosis is free; then one fixed figure in writing Parcels arrive from Newbury, Woodley and Theale

An engineer answers, and looking is free
0800 6890668

Match the symptom to the fault.

Different fault? Begin at the triage →

What you are seeing	What is likely wrong	What happens next
The box reports the volume as degraded	One member has failed and the remaining disk is holding the mirror with nothing behind it	Copy off as much as you can, then send both disks
Both disks are the same age, and both are getting slower	The two members do the same number of hours, so they tend to wear out at the same time	Take the box off power now
A new disk went in, and the data on the good disk was lost	The rebuild synced the wrong way, writing the failing member's copy onto the good disk	Stop there, and send both disks
Files are disappearing from both members at the same time	A bad write is written to both sides of the mirror as it occurs	Do not let it re-sync
The amber light has been on since the spring	The mirror has run on one disk for months with no second copy behind it	Send both, not only the failed one
Windows shows the mirror as Failed Redundancy	A member has been removed from the set. The disks themselves are often fine	Do not re-enable it or import it

Packing and posting: first, get a quote, either over the phone or using the form. Second, add enough packing material to stop the contents moving. Third, put a reasonable value on the parcel, one that represents what your files are worth to you. The parcel should then go tracked and insured to the Oxford intake lab. Alternatively, you can drop it off at the lab yourself, Monday to Friday, 9am to 5:30pm, with no appointment needed. Return postage is covered by us. If you would like one of our engineers to help with the packing before you seal the box, ring us. See our packing and postage page for further details.

What a RAID 1 mirror protects you from, and what it does not.

Two-bay boxes set to mirrorA Synology, QNAP or desktop enclosure holding two disks, with the safer of the two configuration options selected during setup. Since the mirror is described by Linux software RAID on the members themselves, the enclosure can stay on your desk: send the disks, each labelled with its bay, and the network box comes whole only if the disks cannot be taken out.

Boot pairs and RAID cardsThe pair of disks a small server starts from, mirrored either via a hardware controller or using Intel RST on the motherboard. The card writes its own record of the set onto both disks, so this can be read back from images without the original card or the original machine.

Windows, mdadm and ZFS mirrorsA mirrored dynamic volume or Storage Spaces in Windows; an mdadm pair in Linux; a two-disk ZFS or btrfs mirror on a NAS. They all keep their layouts on the disks, all can be built from copies, but all have different methods to decide which member they trust. And that is what you need to know as soon as something goes wrong with them.

Why a rebuild can make it worseRecovery on a mirror is done from image files, where a wrong turn costs nothing but time. That is different from a RAID 5 or RAID 6 rebuild, which uses parity to work out what was on the failed drive. Mirrors hold no parity, so no calculation is done when a disk is rebuilt. A mirror simply copies everything from one disk onto the other, in whichever direction the controller has settled on, and if it goes the wrong way the only good copy is lost by the end of the day.

From your parcel to your files coming back.

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

A case number is assigned to your device when we receive it. An engineer then finds the real fault (free) before anything else occurs; you will be advised clearly of what will be returned and what will not, along with a single written price. Charges cannot begin until you have read this and accepted.

The look costs you nothingA fixed price, in writingStill nothing to pay

Both members copied on their own

Each disk is connected to a write blocker and an exact copy of the entire contents, start to finish, is made before a single file is opened. The sound member is copied as carefully as the one that stopped, because on a mirror the sound member is half the answer. Weak areas get short bursts and are scheduled last. Nothing is ever written back to what you sent in.

Nothing is written to your disksOne member at a time

The two copies compared

A mirror gives you two different versions of the same volume, and the problem is finding out which one to go with. The set's metadata tracks when each member had its last write; the file systems on both images supply the remainder. Where the two disagree, the readable and the more recent copy wins, block by block. This process is documented, so nothing has to be assumed.

Which side is newer, provedCompared, never assumed

The volume built from the better copy

With the two sides settled, the volume is stacked back over the images and the file system restored on that copy. Then the files (accounts, shared folders, databases, photos) are opened rather than counted, since simply having a name in a list does not prove anything. Whatever returns damaged is written down as damaged.

Each file opened, not countedDamaged ones named as damaged

The list first, then the bill, then the files

You look at the file list first and make a decision about what to do. We will never invoice anything until you have said yes, and in most cases you only pay if we recover your data. All of the information that we recover will be put on new media purchased by us specifically for this job, and our company pays the cost of postage back to you. The case stays open on the bench until you tell us the files open on your own machine.

Nothing billed until you say yesNew media, bought for the jobPostage home is on us

What goes wrong most often

A mirror is not a backup and was never sold as one: it makes two identical copies of everything written, at the same moment. So if something gets deleted or corrupted, or ransomware gets in, that happens on both disks at exactly the same moment. It protects against a single failure of either disk. That is worth protecting against, and it is all it protects against.
Two disks bought together fail together: same batch, same shelf, same hours, same heat. So when you lose your first disk, it often does not take months before the second one starts to go; a degraded mirror is a far shorter grace period than people expect, and the disk still sitting in the bay is not automatically the good one.
The direction of a re-sync is everything: a re-sync simply writes one member across the other, and the controller decides which way round on its own logic. Copying a failing drive over a sound one, or a stale member over a newer one, is where most of the bad cases here began. If your box says it will perform a rebuild, tell it no, then ring us before doing anything else.
Send the disk the box gave up on: typically the member that bailed out will be the one holding the newer data, and there will nearly always be enough of a usable image to be worth creating. Sending one disk and keeping the other halves what can be done, and sometimes rules the job out entirely.

Where the honesty comes into it: storage is all that this bench does. Mirrors are what this page covers (two disks in a network box or desktop enclosure, a pair on a RAID card or motherboard controller, a Windows mirrored volume, an mdadm or ZFS mirror), and they all turn up failed, degraded, re-synced the wrong way or written over. If both sides can be read, the volume is rebuilt over the copies and the folder structure comes back as it was before. But nothing can produce blocks missing from both members: when the same area has failed on both, some files will have holes in them, and the report will list them. Nothing survives being written over either. Encryption needs the key from you. We take no phones and no tablets. This work is from £500 + VAT, quoted in writing after a free look, and in most cases you only pay if we recover your data.

How one of these actually went.

RG · RDG-2026-1996ON FILE ✓

A mirror that had been running on one disk since the spring

A distributor off the Theale estate called about an accounts server that could no longer mount its data volume. Months before, one member of the mirror had dropped out (the amber light was on a chassis behind a door nobody uses), and the other disk had carried all the data by itself ever since and was now beginning to fail. Both disks were imaged separately, the worn one in short passes over two nights. Where blocks wouldn't read from the tired disk, they were taken from the older member; where the older member was out of date, the newer copy won. The stock database and eleven years of delivery notes were opened on the bench before anything was sent back. Three days, parcel to parcel.

Each of the members imaged separately3 days from arrival to return

What helps, and what makes it worse.

First things to do

Power off the box, and leave it that way
Send every member, the failed one included
Put the bay number on each disk as you pull it
Say if a rebuild or re-sync has begun

What makes it worse

Letting the box re-sync onto a fresh disk
Reactivating a mirror Windows has marked Failed
Running repair tools against either member
Sending one disk and keeping the other

Answers before you commit to anything.

One disk still works. Why do you want both of them?

Because when a mirror eventually fails you get two versions of the same volume, and usually they're different. One version might have blocks that the other cannot read; the other might be weeks behind, since it left the set before anybody noticed. With both images on the bench, each block is taken from whichever side reads and whichever side is newer, something you'll never get with a striped set. Get a quote first, then send both, with the bay each sat in written on it, and include the one the box has written off.

The box is offering to rebuild onto a new disk. Should I let it?

No. And this is the mistake that costs people the most. A rebuild copies one member right on top of the other, and the controller makes the choice of direction, not you. When the disk left in the bay is the failing one or the stale one, the rebuild puts its data onto the good copy, and now there's no second copy to fall back on. Take the power off the box. Leave your disks where they are. Let images be taken first. Once there's a copy of everything, nothing you press afterwards can undo those copies.

I deleted a folder and it went from both disks at once. Is a mirror not a backup?

It is not; mirrors are mostly bought for the wrong reason. A mirror copies every write to both members at the same time, so a deletion, a corrupt write or a ransomware attack happens at exactly the same moment on each drive. A mirror protects against one disk failing, and no other kind of loss. Afterwards, recovery can only take in whatever has not been written over since, which is why the box should come off power the same day rather than later in the week.

What do I send, and what does it cost?

RAID and NAS work begins at £500 + VAT, and the look before that is free. One single, fixed amount reaches you in writing before anything chargeable begins. First get a quote by phone or with our quote form. Then send each member of the set with its bay number written on it, including the one that was reported as failed, and tell us everything you have tried to date. From a network box, send the disks labelled by bay if they come out; if they cannot be removed, we will accept the entire unit. Disks coming out of a server or desktop enclosure arrive bare in antistatic bags. No collection service is provided, so post them tracked and insured to the Oxford intake lab, or drop them off at the Oxford intake lab, Monday to Friday, 9am to 5:30pm, no appointment needed.

If this sounds like yours, start here.

Servers that will not boot → Fujitsu laptops → Servers and RAID arrays → Every device we handle →

Nothing gets worse while it sits switched off.

There is no charge for the look. You then receive a list of the files that opened, those that did not, and a fixed amount to complete the job. Keep it switched off until you have received that in writing.

0800 6890668

Get it looked at Posting it to Oxford

Drives, cards and whole machines

Start from the fault

Evidence, footage and reports

What it costs, and what happens