Case record · NAS & RAID · MKD-2025-5103
The Box Is Finished, the Disks Are Not.
A Northampton practice rang with its own triage done. The QNAP fell off the network on Thursday and by Friday would not light: the box is finished, the disks were never the problem
. All four were out, labelled by bay before anything moved
. A leisure-centre tender closed Monday at ten. They wanted a fixed figure and a straight answer to can you actually make Monday
.
Same fault? Start here.
0800 6890668
Without the jargon.
A deadline reorders the work; it never reduces it. Rebuilding straight onto the practice's own disks would have bought back a few hours by betting the tender on a flawless run, and if a member falters mid-read there is nothing to fall back on. The honest way to go fast is to run the copies together. All four members image side by side rather than one after another, and the array's geometry gets worked out from whichever copies land first while the others are still filling. Nothing waits on anything, and nothing is skipped to save time.
The kit this job needed.
See how a job runs here →| Kit | Why it was used | What it gives us |
|---|---|---|
| Atola TaskForce 2 | Copied all four members together, which is where the weekend was actually saved | Copies several array members side by side, so a week of imaging finishes in days |
| PC-3000 Portable PRO | Health and firmware questions put to each disk without moving it off the rack | Vendor-level access in a case small enough to live beside the imagers |
| UFS Explorer RAID Recovery | Tested member order, stripe width and parity rotation against the finished copies | Unpicks a NAS layer by layer instead of treating it as one flat array |
What happened on the bench.
All four reading together, nothing in a queue
The four members were on imagers and copying within an hour of the courier, and they stayed that way to the last sector. Feeding disks through one at a time is how a three-day array turns into a week. Running the set together is the difference between Friday to Monday and Friday to an apology.
Work out the geometry while the copies are still running
The metadata on the first completed images gave the candidates: member order, stripe size, and which way parity walks. Each candidate was tested against real directory structures on the volume, not accepted because a controller usually does it that way. By the time the fourth copy finished, the recipe for assembly was already written down.
Build it in software, and check the tender first
The volume was assembled in software across the four images and its filesystem opened at the first attempt. Checking then followed the practice's priorities rather than the disk's: the tender set — drawings, specification, cost plan, the appendices — was checked first and confirmed clear over the phone before the remaining folders had finished validating.
The outcome.
Verified files were with the practice before Sunday was out, and the tender went in on Monday off their own drawings. The speed came from overlapping the stages and saying plainly where each one stood. It did not come from leaving any of them out.
Other jobs from the casebook.
See also RAID & NAS.
Does that sound like your device?
One lesson runs through the lot: cut the power, let us look for free, then decide with the facts in hand.