I'm not exactly sure what happened, or how...But suddenly my debian
randomly froze, and wouldn't let me load into X anymore...

Little do I know about troubleshooting a linux computer without
furthering the damage, I decided to clear my ntfs partition on another
drive, reinstall debian on there, and restore my files from the
previous broken install.

Now comes the problem...I had a monster folder named 'data' with all my
music, etc...And now it's just a 0kb file with a question mark instead
of the permissions set.

I've tried running e2fsck, but that ran for over 15hours before I
cancelled it. Anyone have any idea what went on here?

I'm running kernel 2.6

On Tue, 2006-07-18 at 12:06 -0700, Trixx wrote:
> I'm not exactly sure what happened, or how...But suddenly my debian
> randomly froze, and wouldn't let me load into X anymore...
>
> Little do I know about troubleshooting a linux computer without
> furthering the damage, I decided to clear my ntfs partition on another
> drive, reinstall debian on there, and restore my files from the
> previous broken install.
>
> Now comes the problem...I had a monster folder named 'data' with all my
> music, etc...And now it's just a 0kb file with a question mark instead
> of the permissions set.
>
> I've tried running e2fsck, but that ran for over 15hours before I
> cancelled it. Anyone have any idea what went on here?
>
> I'm running kernel 2.6
>
I've never had to do this but...
AFAIK what you should do is boot the PC from a LIVE distro such as
knoppix (or Ubuntu) and then (at least) mount your home directory so
that you can take a current backup.

BTW Random freezes are typical of hardware problems.

Hope you still have *all* your original files.

--
Jeremy Boden

On Tue, 18 Jul 2006 12:06:50 -0700, Trixx wrote:

> I'm not exactly sure what happened, or how...But suddenly my debian
> randomly froze, and wouldn't let me load into X anymore...
>
> Little do I know about troubleshooting a linux computer without
> furthering the damage, I decided to clear my ntfs partition on another
> drive, reinstall debian on there, and restore my files from the
> previous broken install.
>
> Now comes the problem...I had a monster folder named 'data' with all my
> music, etc...And now it's just a 0kb file with a question mark instead
> of the permissions set.
>
> I've tried running e2fsck, but that ran for over 15hours before I
> cancelled it. Anyone have any idea what went on here?
>
> I'm running kernel 2.6

I recovered a crashed partition last fall - here's what I did:

1) put in a new disk and installed Elive (obviously you can install
whatever you want - you can also boot a Live CD, but you still need
somewhere to put the recovered data).

2) do not mount the bad disk - this will further corrupt it. do this
instead:

dd if=/dev/hdb2 of=savefileimage conv=noerror conv=sync

to save whatever partitions you want to rescue to disk files. noerror is
needed to keep going past errors and sync is needed to replace bad blocks
with zeroes so that data does not get hopelessly scrambled.

3) now you can loop mount the image file and attempt to rescue the data -
try fsck if you want. The reason it didn't work before was that the media
was damaged - so it could not correct what was there - this may have made
the situation worse - best case: it couldn't write anything anyway, so
it's no worse than it was.

good luck - this worked for me; I was able to retrieve all the data I
needed from two partitions. You may need to use skip and count to tell dd
where to read from if the partition table has been messed up.

On 07/18/2006 02:06 PM, Trixx wrote:
> [...]
> Now comes the problem...I had a monster folder named 'data' with all my
> music, etc...And now it's just a 0kb file with a question mark instead
> of the permissions set.
>
> I've tried running e2fsck, but that ran for over 15hours before I
> cancelled it. Anyone have any idea what went on here?
>
> I'm running kernel 2.6
>

Make sure the partition you are checking is either not mounted
or mounted read-only before you run e2fsck. Running e2fsck on
a read-write partition could destroy it.

Sometimes e2fsck will put lost files into the lost+found
directory of that partition, so after a successful run of
e2fsck, you might look there.

For your future installations you might consider using ext3
which has *much* better error recovery than ext2.

Good luck.

ray wrote:
> dd if=/dev/hdb2 of=savefileimage conv=noerror conv=sync

For best results, use dd_rescue instead of dd for this. It reads the
data a few times and averages it to produce the new image. This might
fix a few bad blocks.

Trixx, after you're done, run badblocks on the bad disk to see if it's
broken at the hardware level. Once you see read errors whipping by, just
stop the scan since it'll take ages. If it's still under shop warranty
(usually one year) you might return it to the shop. Factory warranty is
almost always three years, but you'll have to send it to the shop by
post and cover the shipping *to* the factory yourself. They'll suggest
you use DHL or TNT, but regular post will do and saves you a lot of money.

If it's a Maxtor (quite probable since those break alot ;-)) and you
decide to go for warranty, burn this iso:

http://www.maxtor.com/portal/site/Ma...downloadID=113

Boot with it and check the disk, make sure you write down the diagnostic
code. You'll need it to get a replacement.


Good luck,
Steven