A lot of what reaches us from this part of the county is a small NAS in an office or a back room, holding everything a business has. One disk fails, the unit reports it, and the instinct is to slot in a replacement and let it heal. That instinct is what turns one failure into two.

A rebuild reads every sector of every remaining disk, continuously, for hours. Those disks were bought together, installed together, and have run the same hours at the same temperature. A rebuild is the heaviest workload the array will ever see, applied to drives that are all the same age as the one that just died. A second failure part-way through leaves you with a partially rewritten array rather than an intact one.

So power the unit down and leave it. Don’t replace the disk, don’t let it resynchronize, and don’t reseat anything. If drives are already out, note which bay each one came from — that order matters and reconstructing it later is harder than writing it down now. Then call, and say it’s an array before anything else.