A six-bay QNAP configured as RAID 0. One disk died, and with it went the whole volume — because a stripe holds no parity, no mirror, nothing at all to catch the fall.
← All case files · from £500 + VAT
The unit refused to boot. Among its six drives, one had died and was sounding a steady beep — the note a spindle makes when its motor is fighting a platter stack that has seized, not the grumble of a drive that has simply lost track of itself. With that, the array and every share on it had vanished, and QNAP’s support line had already delivered the verdict it’s always forced to give here: a RAID 0 keeps nothing in reserve, so there was nothing they could offer.
No copy existed anywhere else. Everything the firm worked on lived on that one volume — its projects, its media library, its running paperwork.
Three things came out of the diagnosis, building to the one that matters most:
not a loose lead, not a firmware tantrum — that beep meant a physical, mechanical death.
which is the reason the QNAP wouldn’t so much as admit the array had ever been there, let alone flag it as degraded.
the point people struggle most to take in. Striping cuts every file into pieces and scatters those pieces over all six disks at once, so one lost member doesn’t subtract a sixth of your files — it punches a sixth-sized hole in each and every one. Put six disks in a stripe and you’ve made something six times as likely to fail as a lone drive, and when it goes, it goes completely.
The failed member being everything, it was the first thing into the clean-air environment: a set of matched donor heads fitted, then a patient image on a DeepSpar, healthy tracks captured up front and the troublesome ones worked loose after — the five intact drives, meanwhile, cloned without a hitch. Working only ever from the six copies, never the drives, we recovered the block size and the sequence the members had been written in and rebuilt the stripe into one volume that would read.
From the six images the stripe was rebuilt, and the contents went back to the client on an external drive.
You won’t get a figure out of us at this point, and we’d treat one from anyone else with suspicion. A stripe’s recovery hinges on a single thing: how much of the dead member we can lift off its platters. Read those tracks cleanly with donor heads and the whole lot returns. Where a track won’t read, the damage doesn’t land on a handful of files — it’s smeared thinly across all of them, which is exactly what striping guarantees it will do. The largest files take the worst of it, and a file with a slice missing isn’t so much corrupted as reduced to scraps.
So the truthful thing to say before the disk is opened is a span rather than a single number, and that span only closes in once the failed member is actually on the bench.
Calling RAID 0 a RAID is generous. It’s a speed trick that happens to share the acronym: it shows the biggest capacity figure and tops the benchmarks, and on a fair few NAS boxes it’s just what the machine defaults to if you click through the setup — which is how so many people come to be running it having never once picked it on purpose.
So if there’s a NAS in your office, stop and find out what level it’s genuinely set to. Do it today. Two minutes of checking now outweighs anything we could do for you after the fact.
And should a disk have gone already: shut it down, and whatever you do, don’t let it rebuild.
Drop the drive at our Guildford Business Park reception, or post it to us — it costs nothing to find out what happened. You get a written figure from the fixed bands before any work begins.