Scrubbing Software RAID

[irasnyd]

I've set up a large software RAID5 here on my machine recently. I've been doing quite a bit of reading, and I am wondering if there is a way to do "disk scrubbing." Some of the more expensive hardware RAID controllers support this in hardware, but I'm wondering if there's a way to do it with Linux software RAID.

Basically what I want to do is this: Read every block in the RAID. If there is an error, then recover the block using the redundancy in the RAID5, and recover the sector on the hard disk (or mark it bad). This should help in the case of disk failure, where there could be multiple errors across the disks if they are not scrubbed.

I haven't been able to find anyone saying that it is possible, but I haven't found anyone saying it's not possible either.

Thanks,
irasnyd

[NeddySeagoon]

irasnyd,

Hmm - it appears to be unnesaccary today.
The individual drives will provide bad sector remapping. That means that when a sector goes bad, it will be remapped to one of the spares provided for the purpose. When you run out of spare sectors, and the drive cannot hide bad blocks from the operating system, the disk is about to die.

With IDE drives, you can use smartmontools to keep an eye on the number of bad sectors and how its growing. With SATA, you may need the passthrough patch to libsata so the ioctrls that smartmontools needs actually get to the drives.

Thats why modern hard drives never have any visible bad sectors - they are still there from new but they have been hidden.
That means there are no read operations that any OS can do to discover bad blocks.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.