NAS & RAID · bench notes · PLY-2025-0928
The Noise Came First; the Red Light Came Late.
The red light was the end of the story, not the start of it. A picture framer in Truro had kept the box running since 2018, up on a shelf above the mount cutter where nobody was going to catch it with an elbow
, and there it sat from then on. For two weeks the counter staff kept mentioning a grinding noise coming from behind the compressor
. Ten days on, the share would not open for anybody, and the light on the front turned red and did not change back. WD will replace the enclosure under warranty; what the shop had put on it is not part of that.
Sounds like yours? Talk it through.
0800 6890668
Why it happens.
There is one disk in a single-bay My Cloud, and beside it a small Linux computer that exists to hand that disk out over the network. The shares, the web page, the phone app, the light on the front — all of them are services, and every one of them is reading the same platters. Nothing tails off gently, then. The moment the disk stops answering, the lot goes at once, because none of it ever stood apart from that single disk. Two weeks of grinding was the only notice the shop got.
The kit this job called for.
How the job runs →| The kit | What it did here | Why we keep it |
|---|---|---|
| DeepSpar Disk Imager 4 | Mapped with the donor heads in place before reading a surface | Scores every head before it starts, works one surface at a time, and holds resets, timeouts and power in check |
| PC-3000 Data Extractor | Kept resets and timeouts tight while the tired mechanics ran | The PC-3000 runs the copy through the drive's own firmware |
| UFS Explorer RAID Recovery | Unpicked the stacked Linux volumes so folders came back as folders | Rebuilds the set, then works back through whatever volume layers the NAS added on top |
How it ran.
Take the disk out and assess it alone
Out of the box, the 8TB disk was answering within five minutes. Reading it dependably was another matter: the heads were worn out, and the sound the staff had been living with was those heads missing their track. From outside the enclosure none of this shows. The Linux board registers only that its storage has gone quiet, and that is why shares, dashboard and indicator light all stopped together.
Donor heads in, then measure the cost of waiting
The donor stack went in and the mapping was finished before any real reading began. Two weeks of delay had left its mark on one region of the platter. Every night the till PC had been sending its order folder over to the box, writing onto heads that were already on the way out, and that is the band holding the damage. Nothing in the enclosure is able to call a halt; it goes on writing until a person switches it off.
Take the image apart as Linux volumes, not NTFS
WD builds its storage out of stacked Linux layers, so a Windows utility hunting for NTFS finds nothing it can use. Taken apart layer by layer instead, the image opened into the folders the counter staff could name themselves — the order book, artwork filed by customer, job after job — in place of a pile of numbered files.
How it ended up.
The shop’s files travelled home on new media, with the single gap in them stated plainly rather than glossed over. The damaged band held older artwork scanned years ago; none of that came back, and the report names each of those files. The stripped disk went back in its case and the box went off for its warranty swap.
Pages people open after this.
Other jobs on RAID & NAS.
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.