Devices · Arrays and servers
Arrays are rarely killed by the first fault. They are killed by what follows — a rebuild hauling a tired member through one clean read until it gives out, a rejected disk pushed back in, drives swapped between bays on the chance it helps. Mirroring buys less than people think: both halves went in on the same afternoon and have run identical hours since. Sets arrive from Hull's Old Town offices, from the Grimsby docks, from a haulier's yard at Goole. Everything is done on images, so nothing gets worse after they land.
An engineer will take the details
0800 6890668
| What you see | What the fault is | What to try |
|---|---|---|
| Foreign Configuration Found (Dell PERC) | The disks claim a set the card cannot place | Answer nothing. Image every member. |
| 1786 — Drive Array Recovery Needed (HP) | No redundancy left; a rebuild is queued or part-run | Power it off; image the set |
| 1784 — Drive Array Drive Failure (HP) | One of the members is dead | A replacement disk will not mend this |
| 1788 — Improper Drive Replacement (HP) | Disks have been refitted in the wrong order | Stop. Bay order settles this one. |
| Virtual Drive is Degraded / Offline | Parity is gone from the volume, or the volume itself is | Shut the server down completely |
| 1720 — SMART drive detects imminent failure (HP) | A disk still in the set says it is about to go | Image it the same day |
Each item that arrives gets a case number the day it lands. An engineer looks at it and finds the real fault, and that costs you nothing. You are told in plain terms which files can come off the drive and which cannot. A price follows — one figure, written down. No work starts until you have seen it and agreed to it.
Each disk goes onto imaging hardware built for it, the ones the card had already given up on included. All the work from that point runs on the copies. Nothing at all is written back to what you sent.
The layout itself gives up the running order of the disks, the stripe size and the way parity rotates. The set is then stood up over the images on our own hardware. The controller you sent is not used, and at no point is anything asked to rebuild.
Array up, the file system is repaired next, then the virtual-machine containers, then the databases. Everything is reconciled with a full listing before it goes out.
You are not billed while you are still thinking. The list of what came off the drive goes to you first, and the invoice waits on your word. Files go home on media bought in for the job, with the postage back on us. We keep the job open until you have opened them on your own machine.
How single parity got its reputation: a consumer SATA drive is specified by its maker at one sector it cannot read in every 1014 bits — about one per 12.5TB read off it. A rebuild on a large single-parity set demands far more than that from every disk still standing, without a single stumble, and that is the point at which rebuilds stop. The figure belongs to the manufacturers, not to us, and how far it bites in practice is still an argument.
A second member dropped out under the rebuild that Friday, and the live data went beyond reach. From then on the array stayed off. All four disks were imaged onto our own hardware. The stripe was read back from three clean copies, plus the little the fourth would still give up, and the works kept to schedule on the Monday.
Send it with the others. A disk the card has thrown out will often hold the newest copy of certain stripes, and we take each block from whichever member gave it up most cleanly.
No. Which disk sat where, how big the stripe is and which way parity steps round are all recoverable from the contents of the disks themselves. That is bench analysis, not guesswork.
Both. With the array standing, VMDK and VHDX containers and any database stores are pulled off first. Each is then mounted and opened here — a filename on a listing tells you nothing on its own.
A set of several disks is usually four to seven working days. If you have genuinely stopped trading, say so when you ring. The job jumps the queue, and we fit the intake round your deadline.
Leave it switched off. A failing drive has a limited number of starts in it, and each one spends a little more. The free diagnosis will say which files can be read.