Chris Wedgwood wrote:
> It's a pretty rotten answer really :)
>
> I agree that since most (all?) modern drive transparently remap errors
> until the 'grown' table (or equivalent) is full that if you see errors
> your disk is in terrible shape and has been for some time, but it
> doesn't seem unreasonable for someone to want to mark part of the disk
> bad and have the fs avoid this.
>
> Almost ever other fs I can think of allows this kind of thing.
>
> --cw
You should check out the Bad Block on Disk Drive section at:
http://linas.org/linux/raid.html
Which then links to this LKML discussion about the issue about bad block
capabilities in 2.5:
<http://hypermail.idiosynkrasia.net/linux-kernel/archived/2001/week52/0516.html>
Basically even with ext2/ext3, with the drive failing spectacularly,
it's pretty much game over for that drive. EVMS and LVM may have
addressed this issue where the LVM/EVMS layer is actually handling the
physical disks and can reallocate bad blocks on the fly.
As he mentions many drives are using s.m.a.r.t these days, and
monitoring your devices for that is probably useful.
-Mike
|