Typically, when a HD starts developing bad sectors, it means that it'll die soon and you should toss it. It's possible to low-level scan it with the manufacturer's software and mark the bad sectors so you can continue using it for a while, but eventually more bad sectors will develop and you'll be in the same situation again. This goes on faster and faster until the drive becomes utterly useless.
However... what if, instead of marking the bad sectors, one only partitions known-good space, leaving a large buffer of unused disk space around the defects?
Say... run badblocks, badblocks freaks out at sector whatever at 20% of its scan. You partition the disk from 0% to 15%, then from 25% to the end.
Run badblocks again on the second partition. Badblocks freaks out at 60%. You delete the second partition, and remake it from 25% to 55%. Repartition the rest.
You get the point. You keep doing this until there's no more space on the disk.
This results in much less available space than a simple bad sector scan-and-mark, and several small partitions which may or may not be a pain to manage, but it absolutely ensures that the heads won't stay for long in spaces where magnetic damage has occurred (they obviously can't be prevented from flying over them while looking for other data).
My theory is that by only minimally stressing damaged areas, one should lengthen the remaining lifespan of a damaged hard disk enough to do something useful with it - assuming, of course, that a head crash hasn't occurred, in which case total failure is just behind the corner and there isn't anything we can do.
Do you think this theory is sound?
Edit: I should mention I'm not that much of a miser that I must absolutely squeeze all disks up to their very end at the risk of my data; I ask more for those cases where getting a replacement drive might be a problem - for instance, I'm having trouble sourcing PATA laptop drives in decent condition to restore old laptops.
However... what if, instead of marking the bad sectors, one only partitions known-good space, leaving a large buffer of unused disk space around the defects?
Say... run badblocks, badblocks freaks out at sector whatever at 20% of its scan. You partition the disk from 0% to 15%, then from 25% to the end.
Run badblocks again on the second partition. Badblocks freaks out at 60%. You delete the second partition, and remake it from 25% to 55%. Repartition the rest.
You get the point. You keep doing this until there's no more space on the disk.
This results in much less available space than a simple bad sector scan-and-mark, and several small partitions which may or may not be a pain to manage, but it absolutely ensures that the heads won't stay for long in spaces where magnetic damage has occurred (they obviously can't be prevented from flying over them while looking for other data).
My theory is that by only minimally stressing damaged areas, one should lengthen the remaining lifespan of a damaged hard disk enough to do something useful with it - assuming, of course, that a head crash hasn't occurred, in which case total failure is just behind the corner and there isn't anything we can do.
Do you think this theory is sound?
Edit: I should mention I'm not that much of a miser that I must absolutely squeeze all disks up to their very end at the risk of my data; I ask more for those cases where getting a replacement drive might be a problem - for instance, I'm having trouble sourcing PATA laptop drives in decent condition to restore old laptops.
Last edited:
Facebook
Twitter