Open, and looking at it costs nothing · 9am–5:30pm, Mon to Fri Ring instead, it is quicker: 0115 8220606
NDR Nottingham Data Recovery 0115 8220606 Get my quote
NDR / Devices and media / Server and array recovery

Servers · arrays that have stopped

Server data recovery for Nottingham. The server will not boot, and the disks are usually the healthy part.

Sound drives, failed machinery around them: that is what nearly every server here turns out to be. A card whose idea of the array has drifted away from what is really in the bays, a console prompt with two costly buttons on it, a cache holding writes that never reached a platter before the mains failed. Each member is cloned first, and every repair thereafter is done on the clones, so your controller can stay bolted where it is. What comes back is the whole volume — mail stores, shares, guests, databases and the rest. From £500 + VAT for arrays, NAS boxes and servers, one figure in writing after the free diagnosis. A distributor at Worksop and a bioscience company on Pennyfoot Street in Nottingham have posted in the same chassis.

No files back, no bill — on most jobs Free diagnosis first, then a single price, in writing Parcels come in from Derby, Lincoln and Loughborough

Ask an engineer, and the first look is free
0115 8220606

What the server's symptom points to.

Not yours? Open the fault finder →
What is happeningWhat usually causes itWhat that means for you
The PERC screen: Foreign Configuration DetectedThe members and the card no longer agree about the array, so it stops rather than guess. Nothing has gone yetGive the console no answer at all
1786 on an HP: Drive Array Recovery NeededA rebuild has begun, and may have halted part-way throughSwitch it off. Clones before anything else
Amber lights on the bays after a power cutHalf-finished writes were cut off, and whatever the cache was holding at the time is anyone's guess. The array's own description of itself is intactDo not power it back up
It boots, but the data volume is not thereThe card's configuration, or else the file system sitting above it. In each case the sectors underneath are intactComes back, as a rule
The RAID controller itself has goneBay order, stripe size and the rotation of parity are recorded on every memberYou do not need an identical card
Files copied off, but no guest and no database will startA copy that exists and a copy that is coherent are different things, and only the second sort startsFrom there it is database recovery work
How to pack and post it: pack it tight enough that nothing can shift inside the box, declare a value based on the data and not the hardware, then get it to the intake lab on a tracked service. Postage back to you is covered. Would you rather run the packing past an engineer before you seal it? Ring us first. The detail sits on the packing and postage guide.

The map of the array travels on the members.

PERC and MegaRAID controllersAn open format, DDF, is what both cards put into the closing sectors of a member. The clones surrender it readily, and reassembly can then go ahead with your controller left where it is, bolted into the server.
Smart Array cards from HPHP does it differently. Each member carries RIS structures at its head, and the parity block stays on one member for a run of stripe rows before it moves to the next — delayed parity, which is not the rotation a standard preset assumes. Aim a default preset at that and what you build is a volume the server never held.
A prompt where both answers hurtAnswer this one at your peril. Import is capable of laying stale metadata over stripes that were perfectly sound; Clear takes the description off every disk and keeps nothing to put it back. Have the clones taken first and the prompt has no power left to hurt you.
Writes the cache never flushedThe instant battery-backed memory holds a write, that write is reported done — long before any platter sees it. Cut the mains part-way through a flush and it is the file system itself that breaks, invisibly, with nothing on screen to say so. Reason enough to clone before anything else is tried.

From the parcel arriving to the files going home.

Jobs already finished →
01

Booked in the day it arrives, and the first look is on us Free

Your device picks up a case number the day it arrives, and an engineer then establishes what has genuinely gone wrong — at no cost, and ahead of everything else. Two things come back to you together: a plain account of what can be lifted and what cannot, and one fixed figure in writing. Agree to that figure, or turn it down and pay nothing.

The first look is freeOne fixed figure, in writingNo charge at this point
02

The server stays exactly as it is

Shut the server down first. Drives come out singly after that, each labelled with its bay number before it travels anywhere. Nothing the console asks should be answered. Here, purpose-built imaging hardware takes every member on its own — the ones the card had already refused as well — and not a file is opened until the last clone finishes.

Every member imaged separatelyIncluding the ones the card threw out
03

The array is rebuilt in software

The clones bring the DDF or RIS records with them. Extract those and the geometry is yours: how big a stripe is, which RAID level it was, which bay held which member and, on a parity set, the rotation. With that in hand the array is assembled in software, and a card that has failed is no longer anybody's problem.

The layout read out of the membersYour controller never leaves the machine
04

A clone is not yet a working volume.

Repairs go onto the reassembled clone and never onto the drives you posted, whatever the array was carrying — NTFS or ReFS, ext4 or XFS, a VMFS datastore. Then comes real testing. Guests booted, shares mounted, databases attached, individual files opened and read.

Repairs made on clones onlyMounted and booted to prove it
05

The file list comes before the bill

The file list reaches you before any invoice does. Say yes and it is billed; say no and it is not — and on most jobs, if nothing comes back there is nothing to pay. Whatever comes off goes home on media bought in for your case, carriage at our end. Nothing here is closed until you have opened those files on your own machine.

Nothing is charged until you agree the figureNew media, bought in for your jobWe pay to send it home

What comes in most often

  • No two makes keep the map in the same place — RIS sits at the head of an HP member, while DDF, the format PERC and MegaRAID write, lies down in the very last sectors. Look at the wrong end of a disk and every figure you take off it is a guess.
  • A disk the card discarded still counts — a member rejected weeks back may be the only place certain regions still read, which is why every clone is set against the others, byte for byte. Nothing at all is accepted here on trust.
  • Default presets get Smart Array wrong — the parity block on an HP array stays on one member for a run of stripe rows before it rotates, a delayed-parity scheme that standard tools do not assume, so what those tools hand back is a volume your server never had. Assembly has to follow HP's own scheme.
  • The cache battery gets asked about early — the real state of the file system depends on whether writes held in write-back ever made it onto a platter, and that question is settled before anyone is given an expectation to carry.

What those two buttons actually do: the foreign description is treated by Import as true, and written over whatever configuration your card currently holds. Clear wipes that description from every member, keeping nothing to restore it from. One keypress does either, and the screen says nothing to caution you. Hold full clones and neither answer can take a file off you. On a job of this kind, cloning is not one stage among several; the whole of the rest rests on it. Arrays, NAS boxes and servers begin at £500 + VAT. That figure is fixed in writing once the free look is done, and on most jobs nothing is charged if nothing is recovered.

One of these, from start to finish.

NG · NTG-2026-1966ON THE LOG ✓

The cache battery was flat; the practice database still opened whole

A long power cut had emptied the cache battery. When the power came back the controller put a question to the office that nobody there could safely answer, so it sat unanswered. Not one member disk was worked on directly: all of them were cloned, and the array's own metadata came off those clones, which settled the parity order. The rebuild then ran over the copies, and what opened at the end was the whole practice database, the write the cache had still been holding included.

The database opened in full5 days between arrival and return

What helps, and what does damage.

Worth doing first

  • Photograph whatever the controller displays
  • Label every drive with its bay position
  • Send the whole set, apparently finished ones as well
  • Say what the server ran: SQL, Exchange, guests, shares

What makes it harder

  • Choosing Import, or choosing Clear
  • Fitting a spare and starting a rebuild
  • Flashing new firmware onto the card now
  • Powering the server up before clones exist

Answers before you spend anything.

Our server will not start. Is everything gone?

Rarely. What has failed is usually a layer above the disks themselves — the card's record of the array, or the metadata each member carries, or a file system damaged underneath both. Do two things and the good position you are in now is the position you will still be in when the drives arrive at a bench: leave the power off, and let the console's questions go unanswered.

'Foreign Configuration Detected' — what is it telling me?

The metadata on the members has stopped describing the array your card expects to find. Swapping the card, reseating a member into the wrong bay, or letting a failing controller carry on will each produce that prompt. Halting there is correct behaviour and not a fault. Import writes to your disks, and so does Clear, so have clones taken before either one is pressed. With clones already made, whichever button ends up being pushed, no file of yours is at risk.

Do you need the RAID card as well as the disks?

No. Every member carries an account of the array. PERC and MegaRAID record theirs as DDF, which is a published format, while HP prefers RIS structures of its own. Both read straight out of the clones, after which the array is assembled in software — leaving the card in the machine, and the server free to be rebuilt around it later.

Will our databases and our virtual machines run afterwards?

Usually, though this is two jobs rather than one. Before anything else, the reassembled volume has to yield an intact VMDK, or an intact MDF. Persuading a hypervisor to mount the first, or SQL Server to attach the second, is the other job, and the database recovery page on this site is where it is described.

Start here if the same thing is happening to yours.

A RAID 5 rebuild that went wrong → Mirrored RAID 1 sets → Asus and HP laptops → The full device list →

Switched off, a drive cannot get any worse.

The look costs nothing. What comes back to you is a file list — what opened, what did not — plus one figure to finish the job, in writing, before anything chargeable begins. Most jobs carry no fee at all unless the data comes back. Leave the drive switched off until you have that list.

0115 8220606

Send it for a free look Posting it to Manchester