It's a known problem with patchkit 5 & 6, that should be fixed in the
next patch kit; as it will be a while before this patchkit is released,
HP gives a fix to people that have a support contract.
I will therefore ask for this fix quickly. Thanks to Ann Majeske and Dan
Goetzman for their replies.
My statement about patchkit 5 was wrong: the problem is already there
with this patchkit. Patchkit 3 is ok, though.
> Hello managers,
>
> I have a machine A that mounts by NFS a partition from machine B, that's
> installed with 5.1A patchkit 6. If I transfer files by tar towards this
> NFS partition, the setuid bit is lost:
>
> A_at_root:/tmp# ls -al foo
> -rwsr-xr-x 1 root system 12 Feb 11 10:24 foo*
> A_at_root:/tmp# df -k /bar
> Filesystem 1024-blocks Used Available Capacity Mounted on
> B:/bar 6622728 2710799 1772088 61% /bar
> A_at_root:/tmp# tar cf - foo|(cd /bar; tar xfBp -)
> A_at_root:/tmp# ls -al /bar/foo
> -rwxr-xr-x 1 root system 12 Feb 11 10:24 /bar/foo*
> A_at_root:/tmp#
>
> This problem shows up only when B uses patchkit 6, the copy works fine
> with patchkit 3, 5, or no patchkit at all. The system installed on A,
> and the tar program used (/sbin/tar or GNU tar) are not relevant.
>
> Does anyone know a solution for this, apart from downgrading to
> patchkit 5?
>
> Thanks,
--
Nicolas
Received on Fri Feb 13 2004 - 16:34:00 NZDT