The classic array disaster, on the classic schedule. A company’s RAID 5 dropped a disk; their IT support did the reasonable thing — fitted a replacement and started the rebuild — and at 62% the rebuild crashed out as a second, supposedly healthy disk faltered under the strain. The IT director called this bench at 1pm; the disks were on it by 2pm.
A failed rebuild is the scenario the image-first rule exists for: the five working disks were imaged immediately — including the faltering one, handled gently, weak regions last — so that whatever the weekend held, the array’s exact state was preserved beyond further loss. From the images, the RAID 5 geometry was reconstructed and the array rebuilt virtually, the missing member’s stripes regenerated from parity mathematics, and the file system validated on top. No original disk was asked to survive a second rebuild; the only rebuild that ran was on copies that couldn’t be hurt.
The company had its data back for the start of business on Monday morning — a weekend turnaround on a five-disk array, driven by how fast the disks arrived and how little had been ‘tried’ before they did. That’s the honest moral: the 62% failure was recoverable because nobody forced a third rebuild attempt. Arrays forgive one emergency; they rarely forgive improvisation.
Mid-rebuild failure is the most common serious RAID emergency there is — and the most survivable, if everything stops at the failure. The RAID page explains why the second disk so often follows the first. The first step never changes: a free diagnostic and a fixed written quote before anything is at stake — or call 0131 202 0491 and describe what happened.