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 / RAID 5 arrays and failed rebuilds

RAID 5 · one disk of parity, and no more

RAID 5 recovery, Nottingham. Clone every member before anybody rebuilds. More often than not, that decides it.

When you add parity to your RAID 5 array, you are only protected when you lose one disk. The rebuild (one very large sequential read of the entire disk, from start to finish, asked of disks that came from the same carton and have been at the same cabinet temperature since they came out of the box), however, is typically where you will lose a second disk, much more frequently than to a fault. Price always matters first because no one else publishes it: we provide prices starting at £500 + VAT per array, NAS box or server. The total price for the work, once the free look has taken place, is then written down, and your posted drives go home unaltered. Do not base any conclusions on the front panel alone: failed status can be declared by the controller card within a second. Failed is merely a ruling made by the controller card; it does not diagnose. We receive sets from Nottingham and across the East Midlands: a quality assurance office inside a food processing plant in Worksop, a small software development firm in Chesterfield, and a four-bay set behind the trade counter in Lincoln.

Only pay if we recover your data, 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 a RAID 5 symptom points to.

Not yours? Open the fault finder →
What is happeningWhat usually causes itWhat that means for you
The card says a member has goneOnce your one disk of parity is used up, another failure will take your entire volume with itStop it now, before a second disk goes
A rebuild went half the night before it haltedOne disk of the same age hit a soft patch during its full read: this is how most of them endLeave it powered down. Do not try again
Two members reported failedNearly always one finished disk next to one that is merely slow over a few sectors, rarely a matched pairPost the lot; the clones will tell them apart
A thrown-out disk was forced back inStale blocks now sit in among live stripesTouch nothing further
The new controller finds no arrayWhat exists here is a disagreement about metadata. None of your files have actually moved an inchA routine reassembly job
Recovery software has been pointed at the membersSome writing has happened to the disks remaining in the systemPost them anyway, and tell us which tool ran
How to pack and post it: first get a price quote. This can be done by phone (0115 8220606) or via our form. Next pack the disks so that nothing will move. Insure the parcel for the value of your files, not the drive. Use a tracked service to post it to the intake lab in Manchester. Alternatively you can drop it off Monday to Friday between 9am and 5:30pm without an appointment. We pay for the return journey. Still unsure what should go in the box? Before sealing it, give us a ring and an engineer will let you know. All about packing and posting is on our packing and postage guide.

What a RAID 5 rebuild demands of the disks still in the chassis.

How much a rebuild readsA rebuild has one speed, and that is the maximum possible. Every remaining disk in the chassis is read end to end. Put four 16TB members in and the three survivors owe you approximately 48TB between them, each having run for the same number of hours as the member that failed.
Reading the URE specificationThe error rate given by consumer drive manufacturers is 1 unreadable sector per 1014 bits, or approximately 1 problematic sector per 12.5TB. The enterprise rating is 1015, ten times the margin. Consider the rating as the worst case a manufacturer will allow its name to be associated with, not a countdown. It should also be recognised that many working disks run far beyond it. There is nothing inherently incorrect with the number itself. The frightening graphs developed from this number are another issue altogether.
How the card answersAge is the determining factor. Older controllers will throw away all the rebuilding done so far when they encounter a sector that cannot be completed. More current versions of firmware on both PERC and MegaRAID mark the stripe as "punctured" and continue operating. On Linux-based systems, mdadm will log the bad block and continue as well. Losing a single stripe due to an unreadable sector is an acceptable loss. Losing an entire volume due to an unreadable sector never has been.
Why the next disk followsOne carton, one chassis, one temperature history. Ask for one uninterrupted read through every disk, and whatever disk was silently failing gives out there and then. Here we cover the same ground but at a slower pace, holding back the soft areas until right at the end.

From the parcel arriving to the files going home.

01

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

Your case number is assigned when the item is received. Then an engineer works out what went wrong (no charge). At this point, you receive a basic summary of which parts are recoverable and which are not. You also receive a single written quote for the price of the recovery. Accept the quoted price or turn it down, and there is no obligation to pay anything.

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

Clone every member first

Every member has an image made of it on our own storage, including all the members the card wrote off. Every read that takes longer than normal is recorded at the time. The slowest areas are processed last. No bytes go back to the disks you posted.

Each member imaged firstIncluding the ones the card threw out
03

The layout taken off the disks

Before we can assemble anything, we have to work out which bay each disk was in originally, how big the stripe is, and which way parity rotates. All of these answers come from what the disks recorded about themselves, compared with the layout of your data. We do not make assumptions during this process.

Bay order and stripe size worked outConfirmed before anything is assembled
04

Assemble the array above the clones

Assembling happens on the copies of your disks and not on your real disks. The cleanest clone gives its copy of each block. If all the clones fail to give a block, parity fills in for it. This means that even if a disk was written off by the card, its data can still reach you. Once the file system has been fixed, the volume is mounted and checked. Only then do we release anything.

Each block from its cleanest cloneVolume mounted and checked
05

The file list comes before the bill

A complete file list will be provided before any invoicing. Simply say "yes" or "no" to what we found: yes means it is billed, no means it is not, and in most cases you only pay if we recover your data. What we find is posted to you on media purchased specifically for your case, at no additional cost to you. We consider the process complete once you have opened those files yourself on your own computer.

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

What comes in most often

  • Degraded describes what has already happened. The array carried one disk of cover, and the member that left took it away. From then on the array is as vulnerable as a single unprotected drive, with nothing behind it.
  • It's fear that sells array recovery: no bench does dearer work than this one, and much of what is written about it is meant to frighten you. Here instead you get exactly what the imager found on your own disks, put in writing.
  • A puncture loses one stripe and that stripe alone: if current firmware meets a sector it cannot read, it notes the stripe, records it in its log, and carries on. An older card would give up the rebuild altogether, there and then. Once the array is assembled, every punctured stripe can be listed for you by address.
  • Software arrays on Linux keep records: mdadm superblocks, alongside the bad-block log, can show which disk left the array and approximately when. Hardware cards tell you much less than that.

The arithmetic, done once: a consumer disk is rated at one unrecoverable read error over 1014 bits, or roughly a sector every 12.5TB. Enterprise disks are rated 10 times better, at 1015. A rebuild with four 16TB disks needs around 48TB read without fault from the remaining three. So this is arithmetic, not scaremongering. Against that, disks in service routinely run far past their rating. Both are true, and both point to the same order of work: clones first, then a rebuild, since rebuilding after cloning costs you nothing.

One of these, from start to finish.

NG · NTG-2026-1972ON THE LOG ✓

The year-end still went in on time, with two RAID 5 members gone

19 bad sectors were the whole problem. It looked as though the set had lost two disks at once; in fact one member had dropped out silently months before anyone knew. So when a second member stopped during the rebuild, there was nothing left holding it up. Imaging the pair showed how little on the second member was actually unreadable, and three good members plus one almost-good one worked the stripe out between them. The year-end filing still went in on the day.

19 unreadable sectors, and not a second dead disk6 days between arrival and return

What helps, and what does damage.

Worth doing first

  • Cut power to the chassis as soon as any member drops out
  • Write each disk's bay position on it before removal
  • Include the whole set, written-off members as well
  • Name the controller, and say what has been attempted

What makes it harder

  • Starting a rebuild while a disk is missing
  • Forcing a rejected disk back online from the controller card's menu
  • Aiming off-the-shelf software at the disks
  • Fitting a spare and letting the card initialise

Answers before you spend anything.

Two members are showing as failed on the card. Is the volume gone?

It's pretty unusual. A controller will drop a member for reasons entirely its own: a small number of sectors slightly slow to respond is enough. Being removed from an array does not mean the disk is finished. Before attempting anything, we make sure every member is cloned onto our own storage, and parity held by the other members then supplies whatever a single clone cannot. Two members flagged bad by a card is a typical week around here. Two members that genuinely could not be cloned is rare.

We are running degraded, a member short. Do we rebuild?

Not until we've made clones. A rebuild tells every remaining drive in the machine to read everything it holds, end to end, which is exactly what a tired drive can no longer handle. New parity is then laid on top of the existing parity, and it was the old parity that would have reconstructed your volume for you. Once clones have been made, you may rebuild as often as you like, as nothing done after cloning can reach your clones.

Do the URE figures actually matter?

A URE (unrecoverable read error) is a single sector that the drive's own error correction cannot fix. The drive then stops and reports the sector instead of returning incorrect data. Consumer-grade specifications work out at approximately one URE per 12.5TB read, and enterprise specifications at approximately one per 125TB. View these figures as the least that manufacturers are willing to stand behind, not as a countdown in your calendar. Where firmware is modern, meeting a URE marks the stripe as punctured and the rebuild continues.

Does the controller have to come with the disks?

No, the drives alone will do. Each member keeps a record of how the set was built. Where that record is damaged or unreadable, the shape of your own data tells us the same thing. Reassembly happens in software above the clones, your originals are never written to, and your controller stays firmly bolted where it is.

Start here if the same thing is happening to yours.

Server and RAID work → When the server will not start → The disk is not detected → Mirrored RAID 1 sets → The full device list →

Switched off, a drive cannot get any worse.

The look costs nothing. It gives you a file listing of what opened and what did not, along with the cost of finishing the job, in writing, before anything chargeable begins. In most cases you only pay if we recover your data. Do not switch the drive on until you have that listing.

0115 8220606

Get a quote, then send it Posting it to Manchester