Re: How to recover from crash (urgent for me)
On 9/1/2000 Fam. Engelen wrote:
** what should I do? **
I managed to mount one of my windows disks and copy my home
directories. They seem to copy all-right, but I am worried about the
error messages: a lot of 'permission denied' (but I am root??), and
'attempts to read beyond end of system (or similar)' fail.
first my condolences,
you will have to run fsck on the root filesystem and see how much
ruination remains after words, you should not have remounted it
read-write when its damaged however, that will only make it worse,
making backups of what you can is a good idea however before running
fsck since its repairs are often as bad or worse then the disease (ie
all your files end up in lost+found with names like #49589)
when you have finished trying to get your data off of the READONLY
filesystem then run fsck -n /dev/hda1 (or whatever device your root
filesystem is) and see what it says, that will not actually run the
repairs, but you see what its going to ask to do, then you can run it
again as fsck or fsck -y which will repair every error it finds.
after fsck repairs see whats left of the filesystem, look in
/lost+found and see how much got dumped there, check various
directorys to see if they appear intact, such as /etc /bin /sbin and
so on.
it sounds like you have encountered one of the filesystem corruption
bugs in the 2.2 kernel series (from that end of device error you
mentioned specifically) what kernel are you running? I experienced
massive filesystem corruption under 2.2.13, all on the root
filesystem and the ruination was so bad both times (yes twice, the
second time was right after i finished reinstalling and
reconfiguring) i had to just start over and reinstall, 90% of /etc/
was in lost+found along with /bin and lots of other stuff.
if you have backups you can just restore, if not you will have to
reinstall more then likely, trying to fix this is more difficult I'm
afraid (at least nobody answered my question on if there is a better
way to recover after this when it happend to me)
just a note about kernel 2.2.13 which NOBODY should use anymore IMO,
here is a tidbit from 2.2.14's changelog:
http://www.linux.org.uk/VERSION/relnotes.2214.html
Extfs Fix obscure bitmap and block corruption cases under very high load.
and a excerpt out of a 64000+ byte fsck output log:
Pass 1: Checking inodes, blocks, and sizes
Duplicate blocks found... invoking duplicate block passes.
Pass 1B: Rescan for duplicate/bad blocks
Duplicate/bad block(s) in inode 65: 50375
Duplicate/bad block(s) in inode 66: 50376
Duplicate/bad block(s) in inode 67: 50377
Duplicate/bad block(s) in inode 68: 50378 50379 50380
Duplicate/bad block(s) in inode 69: 50381 50382 50383 50384 50385
Duplicate/bad block(s) in inode 70: 50386 50387 50388 50389 50390
Duplicate/bad block(s) in inode 71: 50391 50392 50393 50394
Duplicate/bad block(s) in inode 72: 50395 50396 50397 50398 50399
50400 50401 50402 50403 50404 50405 50406 50407 50408 50409 50410
50411
Duplicate/bad block(s) in inode 73: 50412 50413 50414 50415 50416
Duplicate/bad block(s) in inode 74: 50417 50418 50419 50420 50421
Duplicate/bad block(s) in inode 75: 50422 50423 50424
[... and MUCH more ...]
all the inodes mentioned got a new home in /lost+found...
this is why fsck is a 4 letter word.
--
Ethan Benson
To obtain my PGP key: http://www.alaska.net/~erbenson/pgp/
Reply to: