Devices · servers & RAID arrays
The first dead disk is rarely what finishes an array. The damage comes afterwards — a RAID 5 rebuild flogging a worn survivor until it gives up, a member forced back online, disks shuffled between bays to see. A mirror is no safer: both halves have done identical work since the day they went in. We work on images only, so nothing worsens once your set is here.
Run the problem past an engineer
0800 6890668
| What appears | What's behind it | First move |
|---|---|---|
| Foreign Configuration Found (Dell PERC) | What the card remembers and what the disks say no longer agree | Press neither button; image first |
| 1786 — Drive Array Recovery Needed (HP) | Redundancy has gone: a rebuild is queued, or stopped part-way | Power down first, image second |
| 1784 — Drive Array Drive Failure (HP) | One of the member disks is dead | No guesswork with replacement drives |
| 1788 — Drive Array Reports Incorrect Drive Replacement (HP) | The disks have gone back into the wrong bays | Order is everything — stop here |
| Virtual Drive is Degraded / Offline | The volume is running without cover, or has gone altogether | Switch the whole set off |
| 1720 — SMART drive detects imminent failure (HP) | A disk that still works is warning you it won't for long | Image it the same day |
Every item gets its own case number on the day it lands here. An engineer traces the fault, gives you a plain assessment of what is realistically recoverable, and puts one fixed price in writing. The diagnosis is free, you are under no obligation, and chargeable work starts only when you say go.
Each disk goes onto a purpose-built imaging rig, the ones your controller threw out included. Everything after that is done against the copies; nothing is ever written back to the drives you sent.
Block size, disk order and the direction parity travels are all read back out of the data. Assembly then happens across the images, on our own machines. Your controller takes no part in it, and no rebuild is ever started.
With the array standing again, the file system, the VM containers and the databases get put right, and all of it is checked against a full listing before anything leaves.
You get the full list of recovered files first, and nothing is charged until you approve it. Data goes back on fresh media, return postage paid, and the job stays open on our side until you confirm every file opens at your end.
Where RAID 5's reputation comes from: the spec sheet on a consumer SATA disk allows about one unreadable sector per 1014 bits read — roughly one in every 12.5TB. Rebuilding a large single-parity array asks each survivor to deliver more than that in a single unbroken pass, which is how rebuilds fall over. That figure is published by the makers, not invented by engineers, though whether it bites as hard in practice is still argued over.
A rebuild was running on the Friday when a second member failed, locking up the server's live data. All four disks were imaged; the stripe went back together from three clean images plus part of a fourth, and the factory started Monday as normal.
Yes, send them all. A disk marked 'failed' can still carry the newest version of certain stripes, and during the rebuild we take each block from whichever member read it cleanest.
No. Parity rotation, block size and disk order all fall out of the patterns held in the data. That's standard analysis, not a guess.
Yes. With the array rebuilt, VMDK and VHDX files, plus the database stores, come out first — and every one is mounted to prove it works, rather than simply listed as found.
Reckon on four to seven working days for a multi-drive job. Where an outage has genuinely halted the business, the recovery can be kept running — say so on the phone and the intake is built around it.
Switch a failing device on again and you lose a little more. Open a case before you do — the diagnosis is free, whatever it finds.