During routine maintenance the array was initialised. The RAID metadata was overwritten and DiskStation Manager stopped recognising twelve disks' worth of virtual machines and databases. It was recoverable — and the reason why is the most useful thing on this page.
← All case files · from £500 + VAT
An IT misconfiguration during RAID maintenance. Somebody initialised a twelve-disk RAID 10 array. DSM came back reporting no configuration — the logical structure was gone, and with it the visibility of the virtual machines, databases and operational documents the business ran on. There was no backup.
It is the most dangerous word in any NAS or RAID interface, and it sits in the menu next to entirely harmless ones.
But here is why this job was recoverable, and it is worth understanding: initialising an array typically writes new RAID headers — it does not overwrite the data region. The metadata that describes how the disks fit together is destroyed. The stripes themselves, the actual blocks of your actual files, are still sitting on the platters, untouched.
The map was destroyed. The territory was not. Recovery is therefore a problem of solving the geometry from the surviving data — difficult, but entirely tractable. What makes it unrecoverable is what people do next: letting the NAS complete a full initialisation and resync, which does start writing over the data region.
— DSM could not identify the array at all.
on the disks, fragmented across twelve members.
, so no automatic rebuild could work.
The geometry was solved, the volume reassembled from the clones, and the virtual machines, databases and operational files returned.
A metadata-only overwrite is one of the more recoverable serious failures, precisely because the data region is intact. What decides the outcome is entirely how much writing happened after the mistake. An array that was initialised and immediately powered down is in good shape. One that was left to complete a resync overnight, or had a volume created on it and files copied back, is a different and much worse job.
The moment you realise you have initialised, formatted, or rebuilt the wrong thing: pull the power. Not shut it down gracefully — a graceful shutdown lets queued writes complete. Pull it.
Then leave it off. Every subsequent minute of runtime is the NAS quietly doing exactly what you asked it to do, and the only thing standing between you and your data is how little of that it managed to finish.
And before any maintenance on a live array: photograph the bay order. It costs nothing and it is one of the variables we otherwise have to solve for.
Drop the drive at our Quayside 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.