Andreas Dilger wrote: > > On Feb 17, 2002 16:43 -0500, Joe Krahn wrote: > > When I first installed ext3, it worked as expected. > > After a crash, journals would be played, and no fsck. > > Now, every crash causes an fsck, just like it were > > a regular ext2, even though / has a journal and > > is being mounted as ext3. > > > > Some fs info for / > > Filesystem features: has_journal filetype sparse_super > > Filesystem state: not clean > > Errors behavior: Continue > > The fact that it says "not clean" indicates that it is NOT being mounted > as ext3, because ext3 never does this. Check /proc/mounts to confirm > that it is being mounted as ext2. Yes, /proc/mounts says it is being mounted ext2, even though /bin/mount tells me it is ext3. Now I know where the problem lies. ... > Did you compile ext3 as a module, and/or you are using an initrd? That > is the most common cause for mounting the root as ext2. > This is on a plain, redhat-compiled kernel, which has ext3 as a module. No initrd, using GRUB bootloader. Maybe ext3 can't mount a not-clean fs, so it ends up getting mounted as ext2 for fscking, but then stays that way. I've not changed the init scripts, so maybe this is what happens on RedHat 7.2 whenever you do something to make your root ext3 "not clean", which normally never happens. Does that sound resonable? Thanks, Joe Krahn