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 / Server arrays and RAID sets

Servers · arrays that stopped short

Server data recovery, London. The server will not boot. Nine times in ten the platters never stopped working.

Once they are on the bench, most servers turn out to be healthy platters surrounded by damaged equipment. A controller whose definition of the set does not match the definition of its bays. A console prompt with either of two very costly options. A cache still storing writes that were not stored on any platter because the mains supply had failed. Each member is duplicated prior to anything else being done; everything else is run based on those duplicates, and your controller card will never be removed from the chassis. The volume comes back entire, with all items (shares, mailstores, guests, databases and the rest). Arrays, network-attached storage systems and servers are priced from £500 + VAT, with the free evaluation establishing that price in writing. Traceability records for food processing companies at Park Royal are kept on a small server; a wholesaler on an Enfield estate maintains inventory and accounting records on another. On this bench they look much the same.

Only pay if we recover your data, on most jobs 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

Servers: what the symptom usually means.

Different fault? Try the finder →
What is going onThe usual reason for itWhere that leaves you
PERC, showing Foreign Configuration DetectedThe card and the members no longer tell the same story regarding the set, so it will stop instead of attempting to operate based upon guesses. At this point, nothing has been lost.Leave the prompt unanswered
HP console, 1786: Drive Array Recovery NeededA rebuild was initiated and possibly terminated mid-processSwitch it off at the wall. Clones are created first.
Amber lights along the bays following a loss of powerThe writes were interrupted in mid-process and nobody knows what the cache was still carrying. The set's own description of itself remains intact.Leave the mains off
It boots, and the volume containing data is missingEither the configuration of the card or the file system directly above it. Below both, the sectors remain intactComes back, usually
The controller itself has failed.Each member retains information about what bay it resided in, the stripe size and the location of the parity.An identical card is not required.
Files were copied out, but neither the guests nor the databases would open.There is a difference between a copy existing and a copy being valid (consistent), and only valid copies will open.That becomes database recovery work
Packing it and posting it: get a quote first (by phone or with the form), then pad the box until nothing in it can move and send it to our Guildford intake lab by post, tracked and insured, or drop it off there yourself (Monday to Friday, 9am to 5:30pm) without prior appointment. We cover all return postage costs. Do you wish to speak with an engineer about the packing before you seal the box? Call us first. Full details are on the guide to packing and posting.

Members carry the map of the set.

PERC and MegaRAID cardsBoth cards create DDF, a standard published layout, within the final sectors of each member. The clones provide it freely, and the set is then rebuilt from those clones; your controller will spend the entire time in its designated location.
HP and its Smart Array cardsHP does this differently from the others. RIS structures reside at the beginning of each member, and the parity block resides one stripe clear of where most utilities search for it. If you provide an HP set to a utility using a default preset, you will receive a volume that your server never ran.
A prompt with no safe answerThis is the screen to walk away from. Stale metadata can be written by Import straight onto stripes that were perfectly fine, while Clear removes the metadata from each disk without providing anything to recreate it from. Clone first, and the prompt loses its ability to cause harm.
Writes the cache never got outWhen battery-backed memory accepts a write operation, it reports that operation as completed even though no platter has yet seen it. When the mains supply is interrupted during a flush, it breaks the file system silently, with nothing displayed to indicate such. This is why cloning precedes every other action.

From the box arriving to your files going back.

01

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

As soon as your parcel arrives here, an engineer assigns a case number that same day, prior to determining what has really failed, which is the first thing done and costs you nothing. In addition to a note of what can and cannot be recovered from your drive, you will receive at the same time the fixed, written price for restoring what can be recovered. You can accept this quote, or decline it, which will result in no further obligation on your part and nothing owed to us.

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

Nothing is unbolted from the server

Turn off the machine. Remove each drive individually and mark each with the bay number it occupied before it goes anywhere. Ignore anything the console tells you to do. At this end, dedicated imaging devices work through the individual members one by one, the rejected ones as well, to generate a clone of each drive. Nobody opens a file until all of the clones are generated.

Each member cloned on its ownMembers the controller had dropped included
03

The set is assembled in software

The DDF or RIS records come across with the clones. Review these records and the geometry is defined: RAID level, stripe size, the bay number each member occupied, and the direction of parity flow when present. Once software recombines the volume from this information, controllers that failed cease being relevant.

The geometry, off the membersYour card stays in the chassis
04

A clone is not a working volume yet.

Repairs are made to the recombined clones and never to the drives that were sent, regardless of what was contained on them (NTFS, ReFS, ext4, XFS or a VMFS datastore). Testing occurs on the clones in earnest: guests boot successfully; shares come online; databases connect successfully; single files can be accessed and read completely end-to-end.

Only the clones are repairedBooted and mounted as proof
05

You see the file list before you pay

You will see what has been recovered as a list first, after which a bill exists. Accepting the list means that a bill will be issued. Rejecting the list means that a bill will not be issued, and in most cases you only pay if we recover your data. The recovered data travels from here back to you via new media purchased specifically for your project, with the cost of postage paid by us. Your case is not resolved (i.e., "closed") until you personally view the files on a computer of your own.

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

What arrives most often

  • The layout is not at the same location on every make: HP places RIS structures at the start of each member; DDF, which is how PERC and MegaRAID create layouts, sits at the opposite end. Reading from the wrong end of a member results in figures that are purely speculative.
  • A member that was ejected by the card remains important: a disk that has been dropped for weeks could be the only member that contains readable regions in a specific area; therefore, all clones are compared against each other byte-by-byte. None of this bench-based testing proceeds based upon speculation or assumptions.
  • Preset defaults misinterpret Smart Array: an HP array separates its parity block from where other utilities would normally locate it by one stripe. Use those defaults to assemble an HP array and you receive a volume that your server never ran. Therefore, you need to follow HP's native methodology.
  • The battery for the cache is queried relatively early on: whether write-back was still storing writes that were never accepted by any platter determines whether the file system is valid, and that is established before you know what to anticipate.

What those two answers really do: Import takes the foreign description at face value and overlays it onto whatever configuration the card has. Clear wipes away all descriptions held by each member, with nothing being reserved to rebuild them. Either action is achieved by just one key press, and the screen will warn you about neither. Hold full clones and no reply to the prompt can lose you a file. On work of this kind, cloning is not just one part of the process; everything else in the job continues to operate on top of these clones. The band for arrays, NAS units and servers starts at £500 + VAT, which is written down after completion of the free inspection, and in most cases you only pay if we recover your data.

One job, followed all the way through.

LDN · ADR-2026-3366JOB LOGGED ✓

A cache battery gone flat, and a practice database that opened whole regardless

The cache battery had run out as a result of a prolonged loss of electricity. Once the power returned, the controller requested information from the office that no one present could have answered safely; therefore, it did not receive an answer. No member disk was accessed or used directly. All disks were duplicated (cloned) instead. The array's own metadata was read from the cloned disks and that data was used to establish the parity order. The rebuild process proceeded across all of the duplicate copies, resulting in the full practice database being rebuilt at the completion of the rebuild. At the same time, the write that had remained in the cache was also completed and written into the final copy of the practice database.

That database opened up whole5 days here, and back to you

What helps, and what harms.

Do this much first

  • Take photographs of the controller screen before taking any actions.
  • Put the bay number on each disk.
  • Obtain a quote first. Send all members of the RAID group, including those that appear to be good.
  • Let us know what was running on it: SQL, Exchange, shares, guests.

What sets us back

  • Pressing Import, or pressing Clear
  • Replacing a failed disk with a spare disk and allowing a rebuild to run.
  • Installing a new version of the firmware on the card at this time.
  • Turning the server back on before the clones even exist.

Questions answered before you commit.

Our server stops before the volume shows. Is it all gone?

Rarely. There are typically three items between you and your files, and the lowest item in the list would be the platters. Immediately above them would be the description each member maintains of the set, and above that, the card's own copy of it. Above both would be the file system. Most damage occurs in the upper two items of this list, and both can be reconstructed. There are two practices you can follow to keep yourself in the best possible position: turn off the power supply, and respond to no prompts displayed on the console.

It says Foreign Configuration Detected. What does that mean?

What is happening is that what each member says about itself no longer agrees with what the card expects to see. Either a replacement card was installed, a member was placed in the incorrect bay, or the controller began acting erratically, and any of these would result in prompting you for some input. This halt is proper behaviour and not an error condition. Each reply to the prompt writes information to your disks, therefore please have clones made prior to responding to any prompt on this screen. Once clones exist, nothing anybody does on that screen will lose you a file.

Must the card come too, or only the drives?

Only the disks will be required, each labelled with its bay. Each member contains the description of the set. PERC and MegaRAID cards use DDF, a published layout that any party should be able to read, while HP uses RIS structures of its own. Both formats come out of your clones, and reconstruction occurs using software from there, which means your card can remain in place inside the chassis and you can choose when to rebuild the system around it.

Will the guests and the databases actually run again?

Generally speaking, yes, although consider this two different jobs. Job one concludes once your reassembled volume produces an unbroken VMDK or an unbroken MDF. Job two is mounting that VMDK via a hypervisor or attaching that MDF via SQL Server, and the database recovery page on this site outlines what is involved in doing so.

Begin here if yours is doing the same thing.

When a RAID 5 rebuild fails → A mirrored RAID 1 pair → Toshiba laptop drives → Every device we take →

Nothing gets worse while the power is off.

It doesn't cost you anything for us to look at it. All we do is give you a breakdown of what opened and what did not, plus a single fixed price to complete, written down before you agree to let us continue. On most jobs, you only pay if we recover your data. Please don't turn on the drive until you get that breakdown.

0203 0868633

Get it assessed Sending it to Guildford