[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: fixing a box where the hard disc may failed



Tim:
>> I'm trying to rescue some data from a box where the hard disc *may* have
>> failed.


James Wilkinson
> You know about the smartctl -l error /dev/hda command, don't you? That
> will show you if the *disk* thinks it needs replacing.

Not that particular command, though I am a bit familiar playing with
smart data and hard drives.  I had the smart daemon running, and
configured for that drive, but it hadn't produced any warnings.  That's
one thing that made me suspicious more of a filing system error than
hardware error.  Likewise, the BIOS is supposed to check on smart
details.

> You've got good backups anyway, right?

Not of the couple of bits that I want.  Isn't that always the way?  All
the valuable stuff is backed up in several places.

> It sounds like you really need to reinstall this box.

Yes, that's the intention.  It's really only a test box, hence it
doesn't have much on it.  Unfortunately, it's the last lot of tests that
I want to resume working on, and they're on this box (webserving
experiments).  If I can't, it just means more typing than I want to, to
resume, and remembering how far I got.   The sort of backups I've done
for this, are along the lines of httpd.conf.backup duplicates in a few
places, just not a recent enough one on another box, unfortunately.
Nothing really more valuable than that.  Though this latest minor
disaster is turning into a useful thing in itself - learning to recover
a box, but not when it's vital.  Probably the best time to learn how to
do that.

Current status:  I had a bit of advice from a friend to do a "remount -o
remount,rw /" on it (since, / was read-only despite the mount list
showing it as read & write).  That allowed me to log in, swap a password
file in /etc/ so it's boot *almost* normally.  Now I'll NFS off a few
files I'd like to keep (working nicely), see what else I can break or
fix (not done this yet), then start afresh so it's reliable (tomorrow,
maybe).

So far I've found:  Something screwy with the passwords, so that not all
users details are as they should.  The file looks fine, so there might
be a non-printing character somewhere that I can't see.  GDM won't let
me log-in, but that's no drama.

The hardest thing is going to be a re-install, simply because it has no
floppy or CD-ROM, I'll have to pull the box out of the shelf and open it
up.  That's about as difficult as it gets.  Easy peasy...  I'm so glad
this isn't Windows, and that I'm not faced with registry repairs!

This isn't the worst I've had to fix.  I had a friend get so pissed off
with his Windows box that he threw it about on the concrete floor.
After I resoldered a chip that flew off his graphics card, resoldered
another on his motherboard that was close to falling off, and reseated a
few cards (which, either this, or the chips that came off having dry
joints, would have been the probable cause of the failure), the machine
worked fine, surprisingly (despite all the physical abuse; though he
did, at least, have the sense to remove the hard drive before the
torture).  I was quite amused.

-- 
(Currently running FC4, occasionally trying FC5.)

Don't send private replies to my address, the mailbox is ignored.
I read messages from the public lists.


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]