At 1:10 PM +0200 6/18/06, J.L. Coenders wrote:
>Dear list,
>
>I have a problem.
>Smartd reports that I have bad sectors on my disk. I have tried checking it
>with badblocks which also reports bad sectors.
>I believe I have heard it is possible to repair this in a software manner. Is
>this true, or do I need to buy a new disk?
>
>Btw, fsck does not report any problems.
It is normal for modern large disks to occasionally have some bad sectors.
Most people never check and don't notice them. By default, if the data can
be recovered, the disk automatically remaps that sector (as Mikkel notes:
if that feature is turned on). If the data can't be recovered, the sector
is put on a list to be remapped the next time it is written to. Writing to
bad sectors will "fix" them, but their data is lost.
Many (all) modern disks can automatically scan the disk and remap bad
sectors as they fail. This improves the chances that the sector will still
be readable, and that the remapping will succeed. This is controlled by
SMART and smartctl. It can make the disk a bit slower.
[]# smartctl --offlineauto=on /dev/hdwhatever
SMART will also tell you if the disk has too many bad sectors, or other
problems, though SMART only claims to give a few hours warning of impending
failure.
[]# smartctl --all /dev/hdwhatever
See "man smartctl", and, of course, "man smartd".
One needs backups to recover from disk failures, even just uncorrectable
bad sectors. There must be some tool to map a bad blocks list to the
affected files, but I don't know what it is.
____________________________________________________________________
TonyN.:' <mailto:tonynelson@xxxxxxxxxxxxxxxxx>
' <http://www.georgeanelson.com/>
