Data Recovery Case File · Desktop Externals & Aging Drives · The Closing Window

Mounts occasionally, drops mid-browse, blames him for the ejection: a shrinking duty cycle decoded — why grabbing files by hand through a dying window loses the race, and the metadata-first image that won it

His report described, with unusual clarity, a drive negotiating its own decline. The 750GB external "will not mount on my MacBook — but when I type 'diskutil list' in the terminal, it sees it." Sometimes, though: "occasionally it mounts and I have got a small amount of files off. After a while I have to switch it off as it won't mount at all." And the pattern inside the pattern: "when it does mount and begins to build the contents in the window, after a minute or two it disconnects — and I get the error warning about not disconnecting properly." Work and family photos aboard; happy to talk it through. Three decodes below, and they matter in order: what a drive that mounts occasionally is actually telling you; why the "not ejected properly" scolding is the OS blaming the user for the drive's collapse; and why continuing to rescue files by hand through that window — natural, tempting, partially successful — is the wrong race to run against a countdown that professional imaging runs differently.

The decode: the duty cycle, the false accusation, and the wrong race

What intermittent mounting means: a drive that answers sometimes, briefly, and then needs rest is describing a collapsing duty cycle: it can still perform, but only for a bounded spell before something — overheating electronics, a marginal component, a mechanism drifting out of tolerance as it warms — pushes it past its limit and it drops off the bus. The tell in his account is the trend: each window shorter than the last. That's not random flakiness; it's a budget being spent. diskutil seeing the drive while Finder can't mount it fills in the shape — enumeration is cheap and survives longest, while mounting and sustained reading are the expensive acts that now exceed the budget. A drive on this curve has a finite number of good minutes left in it, and nobody knows the number.

The "not ejected properly" warning, translated: the scolding deserves its decode because it misassigns blame so confidently. The OS shows that warning whenever a mounted volume vanishes without a formal goodbye — it assumes a human yanked the cable. Here, nobody did: the drive itself left the conversation, collapsing mid-browse, and the machine wrote it up as user error. Translated honestly, every one of those warnings reads: "the drive just died again, mid-sentence." They are not nagging; they are timestamps of the collapses — and their increasing frequency is the countdown made visible.

Why hand-copying through the window is the wrong race: his instinct — mount when it deigns to, grab what you can — rescued a small amount, and this archive credits every file brought home. But it's worth being honest about the arithmetic, because it's the page's most useful lesson: manual rescue through a dying window is maximally expensive per file. Browsing makes the drive do its costliest work — populating directory windows, rendering previews, seeking back and forth for whatever the human clicks next — burning the precious minutes on overhead, then collapsing, then requiring rest that may not restore the same budget. Professional imaging runs the identical race with the opposite economics: metadata first — the filesystem's tables and folder maps captured in the first good window, tiny in size, priceless in value — then priority files by map, read in disciplined sequential order with no browsing, no previews, no seeks wasted, through hardware that manages the timeouts and resets so a collapse costs seconds instead of a session. The window is the same; the spend rate is not. When the budget is unknown and shrinking, you buy the map first — then everything after it comes home in order of what matters most.

On the bench

The drive's next window opened on the RapidSpar, and the first minutes went where they buy most: the filesystem metadata — every folder name, every file's address — captured whole before a single photograph was asked for. From that map, the estate came home in priority order: the family photographs and work files first, read sequentially, the inline stabilisation absorbing the drop-offs his MacBook had been suffering — each collapse costing a managed reset rather than a lost session — with the Atola Insight Forensic confirming the condition diagnosis between spells. Window by shrinking window, the 750GB was banked, the last stubborn regions taken on their own patient passes, and the whole was verified by opening and delivered. The improper-ejection warnings ended the only way they ever do: with nothing left that needed the drive to stay.

The outcome

The full estate recovered through managed, priority-ordered imaging — work and family photographs verified and delivered — the shrinking window spent on the map and the cargo instead of the browsing. Free assessment, one fixed written figure including VAT, no recovery, no fee. The doctrine his countdown teaches: a drive that mounts occasionally and for less time each session is spending a finite budget; "not ejected properly" warnings are timestamps of its collapses, not your mistakes; and the race against a closing window is won by economics, not urgency — metadata first, priorities by map, sequential reads through managing hardware. Grab-what-you-can feels like rescue. Imaging is rescue.

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.