A software firm in Reading retired a four-bay NAS during an office move and put the disks in a box, on the reasonable assumption that they could always be plugged in again. A decade later a customer asked for something only that archive held. The disks went back into a spare enclosure and it refused to bring the volume up. There was no configuration file, no note of the order, and nobody left who had set it up.
| Media | Four disks lifted out of a four-bay NAS a decade ago and stored loose since. Returned to an enclosure, they do not present a volume, and no record of the original configuration survives. |
| What was reported | Four NAS disks removed around the time of an office move, kept in storage since, will not bring a volume up when reconnected, no configuration record, no member order known, an archive needed for a customer request. |
| Fault class | Logical, at the array level. The disks themselves were readable. What had gone was the arrangement: member order, block size and parity layout, all of which live in metadata that the enclosure could not agree with. Nothing needed rebuilding on the disks; the set needed reassembling in software. |
| On the bench | Each disk attached separately behind a hardware write blocker on PC-3000 UDMA and profiled, all four imaged on DeepSpar with no member ever asked to rebuild, the array metadata read out of the images, member order and geometry confirmed by testing candidate layouts against filesystem structures rather than by trusting labels, the volume assembled virtually in UFS Explorer and R-Studio, files validated and written to fresh media. |
An array is an arrangement, and the arrangement is the fragile part. Four disks holding fragments of the same files are useless until you know which disk came first, how large each block was and where the parity sat. That information normally lives in metadata the enclosure writes, and enclosures a decade apart in firmware do not always read each other's.
Putting them into a spare enclosure was the risk, not the storage. A NAS that does not recognise a set will often offer to initialise it or start a rebuild, and either of those writes to the disks. They were lucky: the box declined to do anything, and they stopped there instead of clicking through.
Ten years in a box does nothing to a disk. Magnetic storage does not fade on a shelf on any timescale a business cares about. All four spun up, all four readied, and all four read at full speed. The archive had been intact the whole time; it was simply unreadable in the form it was in.
Reassembly is done on copies, and it is done by testing. Once each member is imaged, candidate layouts can be tried against the images as often as needed. The right one is the one where directory structures and file signatures line up across the stripe, which is a matter of proof rather than of guessing, and no attempt can damage anything because the originals are not involved.
Each disk was attached on its own, behind a hardware write blocker, and profiled before anything else. All four were healthy, with modest defect lists and nothing mechanical to report. They were imaged one at a time on DeepSpar, and at no point was any of them put back into an enclosure or allowed to start a rebuild.
The array metadata was then read out of the images. It gave a partial picture, so the remainder was established by testing candidate member orders and block sizes against the filesystem: the correct layout is the one where directory entries resolve and file signatures survive across stripe boundaries. With that fixed, the volume was assembled virtually in UFS Explorer with R-Studio used as a second reading, and the archive was validated before being written to fresh media.
The whole archive came back, including the project the customer had asked about and a decade of correspondence nobody had expected to see again. It went back as an ordinary folder tree on a single drive, so that the next person to need it does not need an enclosure to read it.
Multi-disk sets are quoted from £500 + VAT, which is what this was: four members, imaged individually, reassembled in software. The work was logical from start to finish, with no opening and no donor parts, so the guarantee applied in full and no fix, no fee held.
Do not put them into a different enclosure to see what happens. A box that does not recognise a set will frequently offer to initialise it, and some models start a rebuild without asking clearly, which overwrites the parity you need. If a rebuild has already begun, pull the power rather than let it finish. Label the bay order if you still can, though it is recoverable without labels. Send every member, including any disk you believe has failed, because parity is calculated across the whole set and a set is easier to reassemble complete. Nothing here is booted, and nothing is written to a member disk at any stage.
These files are written up from genuine enquiries taken from households and businesses in Bracknell, across Berkshire and Surrey and into London, with names, firms and anything else identifying stripped out. Each one sets down how the fault was reasoned through, the method it actually needed and the equipment it was worked on. One of them is a job that was turned away, which belongs here as much as the rest of them.
Examining it is free and takes 2 working days from arrival, the figure goes to you in writing and does not move afterwards, and a logical fault carries no fix, no fee. Start online or ring the freephone.