Re: new kernels and disk crash!

[Mouse <mouse%Rodents-Montreal.ORG@localhost>]

>> But not necessarily totally busted.  "Read retries exhausted" sounds
>> to me like a (now-)bad sector, not a whole-drive failure.
> to me too.. I did run fsck -fy and let it go... it missed lots and
> lots of sectors, which seemed contiguous.

You were lucky.  fsck -fy doesn't read data blocks for plain files,
which for most filesystems is most of the used space.  Copying the raw
disk to /dev/null (eg, dd if=/dev/rsd0c of=/dev/null bs=1048576) is a
better way of doing a read test for most purposes.

> Once mounted, i tried to copy some directories to salvage things, but
> tar barfed either about in the same area, or in another are... so
> like "two holes".

Or possibly one large one, with the space in between untouched by tar
for reasons I can only guess at (data blocks belonging to inodes for
which the inodes couldn't be read, for example).

> Luckily I found a replacement disk in my stuff... but the supply is
> getting scarce.

Yeah, I know the feeling. :(

> I wonder how a disk gets bad sectors when just sitting there,

I'm not sure myself; I don't know nearly as much as I'd like about the
physical layer of most disks.  Depending on what's wrong, a reformat (a
true reformat, not the "just rebuild filesystem data structures" that
Windows has redefined the term to mean for far too many people) might
fix it.

Or it might not.  Based on my limited experience, there is _some_
failure mode that appears to be largely age-based and can't be
reformatted out.  Presumably something goes wrong with the magnetic
coating....

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse%rodents-montreal.org@localhost
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B