Data Recovery Case File · Power Users · The Pause Before ddrescue

Good tools, wrong patient: the slow-decline arc decoded, the imaging-software question answered with respect — and a clicking Toshiba recovered in the order the hardware demanded

The enquiry read like a well-kept engineering log. His Toshiba 1TB 2.5″ internal, NTFS: "its performance started to degrade until it became unusable" — no drop, no shock, a pure decline. Present state: "it spins up when powered, but I can hear some repeated clicking sound from the drive heads. It is recognised by Windows, but not accessible/cannot be browsed." And then the line that earns this page its title: "For the moment, I have not made any recovery attempt with ddrescue/hddsuperclone from Linux." A user who knows the serious imaging tools by name, has them ready, and paused — writing to ask before running them. The pause was the most expert act in the enquiry, and it deserves the full honest answer about where those genuinely good tools fit, and where a clicking drive falls on that map.

Two decodes: the slow-decline arc, and the honest map of where imaging tools belong

The arc first, because "degraded until unusable" is a diagnosis in itself: a drive doesn't slow down the way software does. Growing sluggishness is the sound of the drive's internal retry economy inflating — marginal sectors needing second and third reads, the firmware relocating what it can, every operation paying a rising tax — until the taxes exceed the income and "slow" becomes "unusable." His clicking is that arc's final chapter: heads now failing reads outright, the mechanism hunting; and "recognised but not browsable" completes the standard portrait — identity records cheap to serve, substantive reads failing everywhere that matters. A pure-wear decline with no accident to blame is simply a drive completing its lifespan in public. Now the tools, answered with the respect they've earned: ddrescue and its kin are excellent software — this page will not pretend otherwise — built on exactly the right philosophy: image everything readable first, skip errors rather than grinding them, return for the difficult territory last, never write to the source. On a healthy drive with logical problems — deleted files, corrupted filesystems, failing-but-stable media caught early — that philosophy, in those tools, is genuinely the right first move, and plenty of data has been honourably saved by it. But the philosophy has a floor, and his drive had fallen through it: software of any quality can only ask the drive's own hardware to do the reading — and when the heads themselves are the failure (his clicking says they are), every pass, however gently scheduled, flies damaged heads over the platters for hours, spending the mechanical margin the recovery needs. The honest map: logical fault, stable hardware → his tools, correctly; audible mechanical fault → hardware first — donor heads under laminar flow conditions — after which the imaging that follows at the bench runs on the same skip-manage-return philosophy his tools embody, executed with firmware-level and hardware-level control no software can reach. His pause, in other words, wasn't hesitancy. It was a power user correctly recognising that the clicking had moved his case off his tools' map — and stopping precisely at the boundary.

The recovery — his philosophy, one layer deeper

Under laminar airflow the arc's conclusion was confirmed — a head assembly worn to failure, platters carrying the honest wear-pattern of a long decline but clean of trauma — and a matched donor stack restored the mechanism the software route could never have repaired. The PC-3000 then imaged the terabyte running exactly the doctrine his held-in-reserve tools were built on, one layer deeper: cooperative territory banked first at pace, the degraded regions his slowdown had been mapping for months negotiated last in bounded, managed passes, retries governed at the hardware level rather than requested through it. Coverage closed in the high ninety-nines; the NTFS volume mounted whole from the image; his data came back verified across the set on new media — and the report was written for the reader it had: fault, method and coverage stated in full technical detail, the imaging log summarised the way an engineer would want it, and the closing note giving his restraint its due in writing, because a pause that preserves a drive's remaining margin is a contribution to the recovery and should be invoiced to nobody.

Outcome

Full recovery, hardware-first — and the map filed for every capable user with good tools and a misbehaving drive. Let the symptoms route the case: logical faults on quiet, stable hardware are legitimate territory for serious imaging software, used read-only, to a separate destination, stopped at the first sign of distress; audible mechanical symptoms — clicking, scraping, straining — take every software tool off the table at once, however excellent, because the reading hardware is the patient. Respect the tools by respecting their floor: the professional bench doesn't replace the skip-and-return philosophy; it rescues it, by repairing the hardware that philosophy must run on. And keep his habit: the sentence "I have not yet run anything — should I?" is the most valuable one a power user can send, and the answer, for a clicking drive, is the one he suspected: no — and thank you for asking first. The pause was the expertise. The bench just finished what it protected.

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.