[rescue] Repairing bad blocks

[Sheldon T. Hall]

On Wed, 22 May 2002 20:20:36 +0200 Jochen Kunz <jkunz at unixag-kl.fh-kl.de>
wrote:

>On 2002.05.20 19:55 Sheldon T. Hall wrote:
>
>> Over the weekend, I started getting tons of bad-block error messages
>> on the drive in my LX.
>
>Replace the disk _soon_.

Oh, yeah.  I did that.  I was just hoping to back it up one more time before
I did.  The constandre-tries on the bad block made that pretty nearly
impossible, though.

In the event, I swapped in a new machine (replacing the LX with a Classic)
with a larger disk (4.5G vs. 2G), and did a new install from my new CD drive
(20x vs 2x).

Then I put the old disk in a 411 case, mounted the whole thing as /a on the
new disk, and started copying files.  Mounting the old disk RO meant that
the bad blocks, to which the system was trying to write when that disk was a
boot disk, didn't matter.

I'd have loved to just restore from my backup tapes, but the most recent two
weren't readable, and the fscking sysadmin (me) hadn't been following the
backup schedule.  Or testing the tapes.  Bad sysadmin, no banana.

> Maybe you can repair it with sformat, but make frequent backups...
> ftp://ftp.fokus.gmd.de/pub/unix/sformat

I'll get that anyway.  You can't have too many tools!

As a side note, I dearly wish Unix's organization of files made a little
more sense.  Were I king, I'd have it this way:

    stuff that never changes in one place (binaries, libraries, etc.)
    hardware-dependent stuff in one place (vfstab and such)
    site-dependent stuff in one place
    user stuff in one place

It would certainly make backups and restores easier. As it is, places like
/etc are a mix, and you can't just restore *.* unless the machine you're
restoring to is _exactly_ the same as the one you backed up from.

But I digress.

Thanks.

-Shel
--
Sheldon T. Hall
shel at cmhc.com
206-842-2858 (Home)
206-780-7971 (Office)