On Fri, Oct 04, 2002 at 09:09:50AM +0200, Seth Mos wrote:
> At 18:05 3-10-2002 -1000, Sidik Isani wrote:
> >Hello -
> >
> > but attempting to mount without read-only still hangs, apparently
> > indefinitely (there's no activity on the machine.) The version
> > of mount is 2.11u, and we're running the CVS kernel. It's a bit
> > scary to think we're going to have to reformat all our filesystems
> > if we need/want to upgrade beyond 2.4.16? Any idea what is going
> > on? Again, xfs_repair now says this is a perfectly consistent
> > filesystem, so this shouldn't be happening.
>
> The only change to the filesystem over history has been a change in the
> logformat. I have a box here that even saw the 2.4.0-test kernels and it
> still reports no problems and mounts just fine.
Same here, actually. It could be a problem in software RAID5 too. I
hope someone with more experience with backtraces than me can decide.
> One thing that does apply is that you _need_ to have a cleanly unmounted
> partition to be able to mount it with a newer kernel. And it is in the same
> time frame as the kernels you are using.
>
> I think something in the log-part of the filesystem may be playing up.
> After cleanly unmounting the filesystem you might try zerooing the log
> using xfs_repair -L
Yup. Without that, I couldn't even mount at all. After xfs_repair -L,
I could still only mount read-only. Mounting read-write caused the
strange hang even though the log has been zeroed.
Be seeing you,
- Sidik
|