NAS & RAID · bench notes · CDF-2025-0928
The Noise Came First; the Light Came Last.
The light on the front was the last thing to change. A framing shop in Newport had run the box since 2018, sat up on the mount cutter where nobody could knock it
, and there it stayed for seven years. For a fortnight the counter staff kept mentioning a grinding sound under the compressor
. A week and a half later the shared folder was closed to everyone, and the front light was red and staying red. WD replaces the enclosure under warranty. What the shop kept on it is no part of that deal.
Does that match yours? Call us.
0800 6890668
Why that happens.
Count the disks and the rest follows. A single-bay My Cloud holds one, with a small Linux computer alongside it whose entire job is to publish that disk to the network. The shares, the phone app, the browser page, the little status light — every one of them is a service running off the same platters. That is why nothing fades gently. The disk stops answering and the whole lot stops in the same instant, because none of it was ever separate. A fortnight of grinding was the only warning the shop was given.
What we used on this one.
Every step, in order →| The tools | What it did here | Why it helps |
|---|---|---|
| DeepSpar Disk Imager 4 | Mapped the new heads first, then worked through one surface at a time | Maps the heads first, works surface by surface, and keeps resets, timeouts and power in hand |
| PC-3000 Data Extractor | Kept resets and timeouts tight while the worn mechanism read | Imaging driven from inside the drive's own firmware, by the PC-3000 |
| UFS Explorer RAID Recovery | Unpicked the Linux volume layers so the folders came back as folders | Not just the RAID: it also unpicks the volume layers a NAS stacks on top |
How it went.
Take it out of the case and judge the disk
On the bench the 8TB disk answered inside a few minutes. Its heads were worn beyond reliable reading, and the grinding had been the sound of them missing. None of that shows from the appliance end. Its Linux knows only that the storage has gone quiet, so the shares, the dashboard and the light failed in one moment.
Fit donor heads, then count what the delay cost
A matched donor stack went in, and no serious reading began until the map was complete. The cost of that fortnight’s wait sat in a single region of the disk. Every night the till PC had kept copying its order folder over, onto heads that were already failing, and the damage is written into that band. Nothing about the box knows to stop. It will go on writing until the plug comes out.
Read the image for what it is: stacked Linux volumes
WD builds its storage out of Linux layers, so a Windows tool hunting for NTFS comes away with nothing worth having. Opened as the layers they really are, the image resolved into the folders the counter knew — the order book, the customer artwork, job by job — rather than a pile of numbered files.
Signing it off.
The shop’s data went home on fresh media, and the single gap in it was set out plainly rather than glossed over. Older artwork scans sat on the damaged band and none of those came back; every missing file is named in the report. The empty disk went into the case, and the case went off for its warranty swap.
Pages people often read next.
The RAID & NAS jobs.
Is that what yours is doing?
Power it down, get it to us, and hold off on any decision until the diagnosis says what can still be read.