Next time this happens, make sure directory "lost+found" exists at the mount-point
and move the damaged file into that directory.  This will keep the bad blocks
"allocated", but put 'em in a place that folks know not to use.

I.e: if the mount point is /opt/local, and the bad file is /opt/local/foo/bar, do:
	> mkdir /opt/local/lost+found
	> mv /opt/local/foo/bar /opt/local/lost+found

This technique usually works, assuming the bad block isn't an i-node.

If it's a bad i-node (an ls on the _parent_ directory usually fails in this case),
move the entire parent directory into lost+found.

Hope this help's,

-S


Austad, Jay wrote:
> 
> Never used badblocks before, but I ran into a problem last week where I had
> bad blocks on a drive on one of my sun machines, but a disk surface check
> run several times turned up nothing.  When accessing certain files I would
> get an I/O error, and a SCSI error that said trouble reading block
> xxxxxxxxxx.  I had to replace the drive to fix the problem.
> 
> Jay