On Sat, 29 Dec 2001, Ralf G. R. Bergs wrote:
> On Sat, 29 Dec 2001 16:08:24 +0100 (CET), Seth Mos wrote:
>
> >It's possible you might had checked out a CVS tree just in between two
> >fixes where it was broken.
>
> Ok, I've now built a new kernel based upon today's CVS source.
>
> Guess what? The problem's still there, but it seems that things have become
> somewhat better. I only received THREE error messages like the following:
At least some progress. I guess you still have some errors in your log. We
will need to find out what is going on here.
Have you tried running a xfs_repair on it? The last time you lost your
root inode. Let's see what got hit this time. Have you also tried updating
your userspace utils?
It seems like you are on debian and probably have other packages then the
1.0.2. Did you compile the userspace utils yourself or did you use the
.deb files which are on the FTP site in the 1.0.2 directory?
> Based upon this findings I think we can for sure rule out a broken kernel
> release. It must be a systematic bug in the XFS(?) code that hasn't been
> fixed
> so far. But how can we find the location?!
Do you have it compiled with quotas, dmapi acls?
Some people have seen some problems with dmapi although I can't remember
what. If you have that switched on you might try turning it off if you are
not using it.
Cheers
|