Can't remember if it was sysv or maybe even older (: There'r've been times when I thought it could have been useful. BTW did you play with the -e flag on tune2fs?
Cheers, Steve On Fri, 08 Sep 2006 15:52:49 +1200 "Davidson, Brett (Managed Services)" <[EMAIL PROTECTED]> wrote: > Didn't know that option, Steve. > > Can't see it in any of the four different system man pages I've looked > at either. > > Was this an old minix option perchance? (Or was it futher back into Sys5 > history?). > > Brett. > > > -----Original Message----- > > From: Steve Holdoway [mailto:[EMAIL PROTECTED] > > Sent: Friday, 8 September 2006 3:21 p.m. > > To: linux-users@it.canterbury.ac.nz > > Subject: Re: Issue with journal corruption on ext3 > > > > Can you still mount -crashed ( -b 32 )? I haven't used it in > > 20 years, but it *might* work. Then you can remove .journal > > and see if that helps. > > > > > > On Fri, 08 Sep 2006 15:00:13 +1200 > > "Davidson, Brett (Managed Services)" <[EMAIL PROTECTED]> wrote: > > > > > No go. You can't remove a journal while the ext3 filesystem > > underneath > > > has errors. > > > Did try the -O ^has_journal and also tried using the -f > > force option. > > > > > > The problem isn't that the superblock is corrupt. > > > > > > It's that some orphaned links exist that normally fsck would tidy up > > > quite nicely. > > > However, as the filesystem is an ext3 type, fsck tries to run the > > > journal first. > > > No go as that's corrupt. > > > Can't remove the journal (via the -O option) as there are underlying > > > filesystem errors. > > > > > > Classic catch-22. > > > > > > Sounds like it's just fodder for the /dev/null device. :-) > > > > > > > > > > IIRC you can treat it as an ext2 partition ( tune2fs <device> > > > > -O ^has_journal switches it off ), and fsck it ( -b 32 always > > > > helps if you have problems ). Then you can mount it as ext2, > > > > and drop the journals ( rm <mount point>/.journal ). Once > > > > you've got it clean, you can reset it to ext3 again with > > > > tune2fs, and mounting it recerates the journals. > > > > > > > > Dunno if that's any use to you. > > > > > > > > Steve > > > > > > > > > > > > On Fri, 08 Sep 2006 13:36:11 +1200 > > > > "Davidson, Brett (Managed Services)" > > <[EMAIL PROTECTED]> wrote: > > > > > > > > > Anyone confronted the issue where a machine lost power > > and both the > > > > > journal and ext filesystem underneath have errors? > > > > > To fix the ext3 filesystem, fsck normally applies the > > > > journal first but > > > > > as that's also corrupt, things don't play. > > > > > > > > > > I have restored the system from the last backup but have > > > > kept a "dd" of > > > > > the filesystem so I can play later as a loopback. > > > > > > > > > > It appears that the journal starts at block 0 and debugfs > > > > won't let you > > > > > touch block 0 so I can't just remove the inode and/or the > > > > journal file. > > > > > > > > > > Should I delete the filesystem copy as there is absolutely > > > > nothgin to be > > > > > done or is there some fun learning to be had? > > > > > > > > > > Brett. > > > > > > > > > > > > > > > Brett Davidson : RHCE, MCSE, SCSA, NZCE(Electronics&Computing), > > > > > TC(Electronics) > > > > > Systems Support Specialist > > > > > HP Christchurch, New Zealand > > > > > > > > > > Phone : +64 3 962 5773 > > > > > Fax : +64 3 962 5747 > > > > > Email : [EMAIL PROTECTED] > > > > > > > > > > > > > > > > > > > > > > > > >
