On Fri, Dec 06, 2002 at 01:37:23AM -0500, Adam Thornton wrote: > On Fri, Dec 06, 2002 at 12:38:55PM +0800, John Summerfield wrote: > > > Okay, use '-f' to force it to check. Just in case. It is telling you the > > dirty bit's not set and that means nobody else has it mounted rw. > > > I have seen e2fsck find errors when forced to check, even though the > > filesystem was supposedly clean. After that I'm out of iudeas. > > Well, it causes a WHOLE BUNCH of the same sort of CCW errors (not > surprisingly). But lots (orders of magnitude) more than when I just mount > it as ext3.
Was that e2fsck -f, or e2fsck -f -n? e2fsck -f may write to the filesystem. > I'm pretty convinced that mounting an ext3 filesystem attempts to write to > the journal inode, even if you want to mount the filesystem r/o. I > consider this a bug. Maybe I'll look at the code tomorrow. It will definitely write to the device if there are things that need to be done (such as replaying the journal), even if it is to be mounted read-only. Can you get read-write access to the device from elsewhere and run e2fsck -f on it? > I also wonder what happens if I set the device node to be r/o. The permission bits are not significant here. -- - mdz
