An eight-drive HP ProLiant array that had quietly been running on borrowed time. One degraded disk is what RAID 5 is designed for. Two is what it is not.
← All case files · from £500 + VAT
A corporate ProLiant that became inaccessible. The controller could no longer read parts of the array, HP's own support had diagnosed degraded drives and stopped there, and the server held financial records, client databases and operational files.
A disk that has developed bad sectors does not announce itself. Parity quietly covers for it, the array keeps serving files, and everything looks normal — for months. The array is now running with no redundancy at all, and nobody knows.
Then a second disk develops a problem, and the whole thing stops at once. It feels like a sudden catastrophic failure. It is nothing of the sort: it is the second failure of two, and the first one happened a long time ago.
With eight members, the odds of that are not small. Every disk you add is another chance for the silent first failure.
, which is what had finally taken the controller down.
, so the controller could not identify the correct configuration.
This was the finding that made the job viable — with parity readable, missing blocks can be calculated back rather than guessed at.
The critical question on a job like this is never "have two disks failed". It is "do the two disks' bad sectors overlap in the same stripe". If they do not, every stripe still has enough surviving members to reconstruct. If they do, that stripe is gone and no tool on earth will bring it back. That is established by imaging, not by inspection.
Parity held, the array was reconstructed from the clones, and the databases, financial records and operational files came back.
What decides these jobs is not how many disks are degraded — it is whether the damaged regions overlap. Two failing disks whose bad sectors fall in different stripes is a recoverable array. Two whose bad sectors collide is a partial recovery at best, and there is no technique, tool or budget that changes it. Anyone who quotes you a percentage before imaging all the members is inventing it.
Monitor your array, and believe it when it tells you something. A degraded RAID 5 is not a warning to deal with next quarter. It is an array with no redundancy left, and it is one bad sector away from being a forensic job.
Do not rebuild onto a fresh disk when a second member is already flaky. A rebuild reads every sector of every remaining disk — the heaviest load those drives have ever seen, applied to drives of the same age, from the same batch, with the same hours on them as the one that just died. This is why arrays so often lose a second disk during the rebuild.
Image first. Rebuild afterwards, from images, with the originals safe.
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.