Call us — 01865 593000
Mon–Fri · 9am–5:30pm · No fix, no fee
Start a free diagnostic →
← All case files // case file · RAID / Server

RAID 6 tolerates two failures. This one had three.

An eight-disk array lost a second member during a rebuild and a third while the business tried to keep trading on it. By the time it reached us the volume was gone and the remaining disks had been running hard for four days.

Device8×6 TB RAID 6
FaultThree members lost
Turnaround9 days
Outcome99% recovered
ToolsDeepSpar DDI · array analysis

The situation

A manufacturing business ran its production and design files on an eight-disk RAID 6. One disk had failed months earlier and been left; a second failed during the replacement rebuild, and the array degraded to its limit. Rather than stopping, the rebuild was restarted twice more while staff continued working from the volume — and a third member dropped out on the second attempt.

First look and diagnosis

RAID 6 carries two parity blocks per stripe and genuinely survives two failures. Three is beyond the design, but not automatically beyond recovery: the question is whether the third disk failed completely or simply dropped out. Here the third was electrically sound with a large but bounded band of bad sectors, and the second was mechanically failing. Only the first was truly gone.

The recovery

All eight members were imaged read-only, worst-condition first — the disks least likely to survive were read while they still responded, rather than being left until last. The mechanically failing member needed a donor head-stack before it would image at all. Unreadable sectors were logged rather than guessed at, so the map of gaps was known precisely.

The array geometry was then derived from the images and the RAID 6 assembled virtually. With six clean members plus substantial coverage from the seventh and eighth, dual parity reconstructed the overwhelming majority of stripes — only those with three unreadable members in the same place could not be resolved.

Verifying and returning the data

CAD files were opened in their application, production databases checked for structural consistency rather than merely copied, and the folder structure matched against what the business expected. The small number of unresolved files were reported by name.

Outcome

Around 99% recovered. Worth noting what actually caused the loss: not the first failure, which RAID 6 absorbed easily, but four days of continued use and three rebuild attempts on a degraded array. Had it been powered down when the second disk dropped, this would have been a straightforward job.

The rebuild attempts caused more damage than the failures

A rebuild reads every sector of every surviving member — hours of maximum stress on disks the same age as the ones that just failed. Repeating it on an already-degraded array is how a recoverable set becomes a marginal one. When an array passes its redundancy limit, power it down and preserve bay order. Emergency RAID work starts from £500 +VAT.

// sending your device in

Two simple steps.

Send us your device for a free diagnostic, and tell us a little about what happened — an engineer will review it and confirm your exact quote in writing before any work begins.

1

Send us your device

Getting your data back begins with getting the device to us. Pack it up safely, pop your contact details inside, and send it over — once we’ve run the free diagnostic, we’ll confirm your exact price in writing before any work starts.

How to pack it
  • Box the device up in a small, sturdy carton or a padded envelope.
  • You can leave out caddies, cables and power supplies — none of them are needed for the recovery.
  • Pop your details inside — name, address, phone and email, on a slip of paper or via our shipping form — and seal it up.
Post toOxford Data Recovery
John Eccles House, Oxford Science Park
Oxford OX2
Shipping formPDF · print & include with your devicePDF ↓

Posting it? A tracked, insured service is what we’d recommend. Rather drop it in? You’re welcome Monday to Friday, 9am to 5:30pm — just package the device up as above first.

2

Need more information?

Want a bit more detail first? Fill in the form with more about your issue and an engineer will review it and send you a custom quote.

An engineer reviews every enquiry personally — we usually reply within 30 minutes during the day. Prefer to call? 01865 593000.

Thanks — your message is in.

We’ll be in touch shortly. For anything urgent, call 01865 593000.

Common questions

Can a RAID 6 be recovered after three disks fail?

Often, yes. Three failures exceed the design but not necessarily the recovery — what matters is whether the third disk failed completely or merely dropped out. A member with bad sectors still contributes most of its data, and dual parity fills a great deal of the rest.

Should I keep retrying the rebuild?

No. Each attempt reads every sector of every survivor, and on disks the same age as the ones that already failed, that is the operation most likely to cost you another member. Power the array down instead.

Can we keep working from a degraded array?

It is the most expensive decision available. A degraded array has no margin left, and continued use is the commonest reason a recoverable set becomes a partial one. Stop, secure the data, then rebuild.

What does emergency RAID recovery cost?

From £500 plus VAT, quoted in writing after a free 48-hour diagnostic, with priority handling available. Physical drive work carries a 50% deposit and the balance is due only on success.

Related

// ready when you are

Facing something similar? Let's help.

Start with an instant online quote, or call and talk it through with us first. You'll have a clear, fixed price before any work begins.

John Eccles House, Littlemore, Oxford OX2 · Mon–Fri 9am–5:30pm · No fix, no fee on most jobs