NAS & RAID · bench file · BHD-2025-8928
The Hum Was the Warning; Red Came Later.
A Coventry framing shop had run the box since 2018, on top of the mount cutter, out of everyone’s way
, and out of the way is where it stayed for seven years. In the second week the counter staff started mentioning a sort of grinding under the compressor noise
. Ten days after that, nobody could get into the shared folder and the front light had gone to solid red. WD will replace the enclosure under warranty. It has never replaced what the shop kept on it.
Sounds like yours? Give us a ring.
0800 6890668
What that means.
The word on the badge does a lot of harm here. A single-bay My Cloud is a hard disk with a small Linux computer bolted to it, and that computer’s only job is to hand the disk out over the network. Shares, phone app, web page, status light: all of it rests on one platter stack. So there is no partial failure. The disk stops answering and every one of those services stops in the same second, because they were never separate things. Two weeks of grinding was the disk telling the shop what was coming.
What did the work here.
The order of the work →| Equipment | The job it did | What it adds |
|---|---|---|
| DeepSpar Disk Imager 4 | Built a fresh head map after the swap and read surface by surface | Head mapping first, then one surface at a time, with resets, timeouts and power all governed |
| PC-3000 Data Extractor | Held timeouts and resets short while the tired mechanism worked | Reads the disk from inside its firmware, using the PC-3000's imager |
| UFS Explorer RAID Recovery | Took apart the box’s layered Linux volumes so folders returned as folders | Reads the volume layers a NAS puts over its array, not only the RAID underneath |
What we did.
Strip the case off and judge the disk itself
Out of the plastic, the 8TB disk gave its answer in minutes. The heads had worn past the point of reading reliably, and the grind was them missing. From the appliance side none of that is visible; its Linux only sees storage that has gone silent, so shares, dashboard and light all failed together.
Fit a donor stack, then find out what the wear cost
Heads came off a matched donor, and nothing was read in earnest until the map was finished. The bill for two weeks of patience showed up in one region: the till PC had gone on copying its order folder across, night after night, onto heads that were already going, and that band of surface carried the marks. Nothing in the box knows how to give up. It will keep writing until someone pulls the plug.
Read the copy as what it is — a Linux volume stack
WD stacks its storage in Linux layers, so Windows tooling looking for NTFS finds nothing worth having. Opened as the layers it actually is, the copy came apart into the folders the counter recognised — the order book, customer artwork, job by job — instead of a heap of numbered files.
How it closed.
The shop got its data back on fresh media, and the one gap was written down rather than glossed. Older artwork scans had been sitting on the damaged region and none of them came back; the report names every file. The enclosure went off for its warranty swap with an empty disk in it.
Other pages worth a look here.
The RAID & NAS cases.
Is yours behaving the same way?
Switch the device off, post it in, and decide nothing until the diagnosis comes back to tell you where you stand.