Portable drives · job notes · PLY-2025-0319
It Slowed for Six Weeks and Then Vanished.
Six weeks was all it took. The volunteer who keeps the records for a village in the Tamar Valley, west of Tavistock, watched the drive give ground week by week — it used to take a minute to show up, and then it was five
, then click a folder and the wheel just goes round
, and finally the lamp lights and nothing else happens
. On it was the village's documentary record: minute books, scanned page by page back to the 1890s, the school log books, and photographs carried in on open days. He had been told once to defragment the drive and never done anything about it, which is the luckiest detail in the case.
Sounds like yours? Talk it through.
0800 6890668
Why it happens.
Slowness of this kind is arithmetic, not fatigue. Nothing inside the case is stiffening with age. A sector that gives itself up on a second attempt costs no time worth counting; a sector that needs fifty attempts costs a lot, and the total climbs as the poor region widens. Previews go first, since building a thumbnail obliges the machine to read an entire file through, retries and all. This is not a drive working more slowly — it is a drive being asked repeatedly for the same sector. And the day it stops appearing altogether is the identical sum done at mount time: the exchange a computer requires before it will list a volume has grown longer than that computer will sit and wait for.
The kit this job called for.
How the job runs →| The kit | What it did here | Why we keep it |
|---|---|---|
| DeepSpar Disk Imager 4 | Ran the awkward passes over SATA, with timeouts chosen to suit the late head | Scores every head before it starts, works one surface at a time, and holds resets, timeouts and power in check |
| DeepSpar USB Imager | Made the opening pass across the enclosure's own bridge, no operating system anywhere in the chain | Images at full speed over the drive's own USB bridge |
| UFS Explorer Professional Recovery | Put the directory structure back together from the images, then wrote it onto the replacement disk | Mounts what other software will not: APFS, ReFS, XFS, ZFS and Btrfs volumes |
How it ran.
Measure before opening anything
Nothing got opened on arrival. There are only so many reads left in a failing drive, and one spent satisfying curiosity is gone for good, so measurement came first. Reallocations stood high and were climbing still, and a survey taken head by head caught one of the four coming back late — later again with each pass over it. The zone that head is responsible for was exactly where previews had given up.
Lift the sound areas while they are still sound
Imaging followed confidence rather than address. Running at full rate, the three dependable heads finished their zones inside a working day, and that put most of the archive beyond the reach of the fault while the drive was still comparatively willing. Only then were the restraints lifted — longer waits, resets under supervision — and the work turned to the area responsible for six weeks of hanging around.
Work the failing head, then rebuild onto a sound disk
The late head's zone was covered slowly, with a stop and a power cycle at each stall. Everything that came off went onto a fresh disk, arranged as the archive had always been arranged, so the cataloguing could be picked up again at the point where it stopped.
How it ended up.
Losses were slight. The minute books returned, so did the school log books, and with them nearly every open-day photograph. One batch from the earliest scanning sessions sat in the territory of the late head: some of it returned damaged, some of it not at all, and the report identifies every affected item. Those volumes can be scanned afresh, because the originals live in the parish safe. The photographs cannot — they were lent for a single day and went back to their families years ago.
Pages people open after this.
Other cases from the pocket-drive file.
Seeing the same thing?
Leave it switched off and post it in. Nothing happens until the diagnosis, which lists what can still be read and what cannot.