I've updated the RHEL4 U5 patch set in the bug to address the
two major issues we observed.  Both of which would have effected
1.6 based releases as much as the 1.4.8 released.  Issues fixed:

vfs_intent-2.6-rhel4.patch:

  RHEL4-U5 introduced upstream vfs_{l,f,}stat64 calls which did
  not tie in to the previous intent hooks.  Added a common
  vfs_getattr64_it() call to be used in these cases.

vfs_nointent-2.6-rhel4.patch:

  Resolves inconsistent i_alloc_sem handling in vfs_nointent-2.6-rhel4.patch
  which was introduced by RHEL linux-2.6.9-fs-lustre-support.patch.  

Thanks,
Brian

> Our patched 1.4.8 which we currently run in production on U3
> (not U4 - I misspoke).
>
> To put a finer point on the problem description:
>
> llmount.sh
> cp /etc/passwd /mnt/lustre
> diff /etc/passwd /mnt/lustre/passwd # looks good
> llmountcleanup.sh
> llrmount.sh
> ls -l /mnt/lustre/passwd # zero length
>
> When I mount the ost image with -o loop I can see the file contents.
> I'm picking apart the mds image now to try to figure what's gone wrong.
>
> Jim
>
> On Fri, 30 Mar 2007, Nathaniel Rutman wrote:
> > Which version of Lustre?
> >
> > Jim Garlick wrote:
> >> Hey has anyone got Lustre running on RHEL4 U5 linux-2.6.9-49.EL?
> >> We munged our patches from U4 and posted them here:
> >>
> >>   https://bugzilla.lustre.org/show_bug.cgi?id=11823
> >>
> >> but we've still got problems - the acceptance-small.sh test fails
> >> when copying /etc into lustre and comparing checksums after
> >> unmount/remount.  All the files seem to be zero length after the
> >> remount.
> >>
> >> Just wondering if anyone else had it working yet - didn't want to
> >> reinvent the wheel.
> >>
> >> Jim
> >>
> >> _______________________________________________
> >> Lustre-discuss mailing list
> >> [email protected]
> >> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss
>
> _______________________________________________
> Lustre-discuss mailing list
> [email protected]
> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss

_______________________________________________
Lustre-discuss mailing list
[email protected]
https://mail.clusterfs.com/mailman/listinfo/lustre-discuss

Reply via email to