Data Recovery Case File · Power Users, Developers & Linux · What Do You Think?

"Fairly confident it's the disk" — confirmed; and of his three suspects, the power button did it: an NVMe locked by its own interrupted housekeeping, unlocked through the service door

The enquiry read like a colleague's incident ticket, which is the highest compliment this archive pays. A Samsung 2TB NVMe in a compact unit: "this morning it just stopped working — the system won't boot. It won't show in the BIOS." The elimination, run properly: reseated; moved to a second identical unit; tried in his laptop's spare slot — "it couldn't be detected by the BIOS at any point, so fairly confident it's the disk." Then the part that makes this case file: a volunteered suspect list. "Possible culprits: hard powering down the PC a few dozen times, either holding the power button or just removing power. Also, the host PC was used in a prototype industrial project and it would suffer a lot of physical shock, maybe some dust." And the closing line — "What do you think?" — an enquiry asking first for a verdict rather than a price. Both follow, in that order: his elimination endorsed in full, and his suspects ranked with the honesty an engineer's question deserves. The short version: the power button did it.

The decode: the verdict he asked for

His elimination, countersigned: three hosts, including an identical unit and an unrelated laptop, all blind at BIOS level after reseating — that is a complete, correctly-run isolation. Firmware-level invisibility everywhere means the drive is failing before any operating system enters the room: its controller never completes its own start-up and announces itself. "Fairly confident it's the disk" was, formally, correct — the bench's only amendment is to remove "fairly."

Suspect one — the power button, convicted: a solid-state drive is never merely storing; its controller runs constant private housekeeping — reorganising cells, updating the internal ledger that maps where every piece of data currently lives. That ledger is the drive's entire ability to be a drive. Cut power mid-update — and a few dozen hard cuts all but guarantee some landed mid-update — and the ledger can be torn badly enough that the controller, on next start, cannot reconstruct its own world. It does the conservative thing: refuses to come up at all. Dead to every BIOS, everywhere, while the memory chips beneath sit full and healthy. This failure signature — sudden, total, unprovoked-looking, after a history of dirty shutdowns — is the classic of the class.

Suspects two and three, largely acquitted: shock and dust are honest suspects for the rig — connectors, sockets, spinning drives if there were any — but the drive on trial has no moving parts to knock out of alignment, and reseating across three hosts already discounted contact contamination. Industrial vibration is not nothing for solder joints over years; it is simply outranked, decisively, by dozens of interrupted housekeeping cycles. The ranking he asked for: power-loss lockup first by a distance, connector-level effects a remote second, shock-as-mechanism essentially dismissed for this component.

The recovery: through the service door

Drives in this state keep a service entrance — manufacturer-level access modes that exist for exactly this rescue. Through it, the controller was brought up on the bench without its torn ledger, the internal translation was reconstructed from the memory's own records, and the drive's contents were imaged in full — the project's estate captured to safety before anything else was contemplated. Verification confirmed what the diagnosis predicted: the data had never been in danger; only the map to it had.

The outcome

Two terabytes delivered verified on new media, with the written verdict he'd actually asked for attached: culprit, mechanism, and ranking, engineer to engineer. Free assessment, one fixed written figure including VAT, no recovery, no fee. And one recommendation for the prototype rig, offered with respect for the environment it works in: give the unit a graceful-shutdown path — a controlled power rail, however the integration allows — because the drive that replaces this one will keep the same housekeeping schedule, and the power button will keep not knowing about it.

Our case files are drawn from genuine enquiries received by our laboratory over the past ten years, anonymised to protect client confidentiality. Each one describes the diagnostic and recovery procedure our engineers apply to that fault, using the equipment listed.