Re: Need help recovering from a power outage [solved]
On Tue, May 10, 2005 at 10:30:10AM +0200, Martin Dickopp wrote:
> Paul E Condon <pecondon@mesanetworks.net> writes:
>
> > I have a Sarge machine on which there is a 60GB HD
> > I use it to run simulations. About 30GB on HD is some
> > useful results, but it has suddenly become somewhat
> > inaccessible because the ext3 journal became corrupted
> > during the outage. By mounting the partition ext2, I
> > can see the data files and they look OK. I think I
> > need to delete the journal, but HOW? All the utils
> > that I have found assume that the journal is surely
> > good, and attempt to run it before checking for any
> > disk corruption. Here, it seems that this is exactly
> > the wrong thing to do. I think I have enough spare
> > space on another HD to copy the files, but it would
> > be so much simpler if I could just zap the journal.
> > Any ideas?
>
> You can remove the journal with tune2fs -O ^has_journal (see the manual
I had already tried this, but there is more that must be done. I tried
again, and forced through...
First one needs to edit /etc/fstab so that mount is not attempted during
boot. (Because failed attempt to read journal leaves partition mounted
and files in use, which makes it impossible to unmount, and above command
won't work on a mounted fs.
Furthermore, even during reboot with noauto, the partition is checked
somehow and the bad journal discovered. So, one must push on ignoring
all error messages, do fsck during boot, answer yes to a bunch of fix
and rewrite prompts, etc.
Apparently, marking the partition ^has_journal does not stop the boot
system from looking for a journal. When it looks, and finds corruption
it forces an fsck before the user(me) is psychologically ready for it.
I would have preferred to mount the partition ext2 and look around
before jumping into an fsck, but maybe that would have been a bad move.
Currently running a diff -r on the partition against a copy on another
computer that I made across the LAN. So far, no differences, but it's
slow.
Thanks for the encouragement to do the awful deed.
--
Paul E Condon
pecondon@mesanetworks.net
Reply to: