wliao@midway.uchicago.edu (Wenchi Liao) writes:
>This is from the man page of restore(8) in 1.4.1 (BUGS section):

>     A level zero dump must be done after a full restore.  Because restore
>     runs in user code, it has no control over inode allocation; thus a full
>     dump must be done to get a new set of directories reflecting the new in-
>     ode numbering, even though the content of the files is unchanged.

>What does this mean? running dump on the restored machine will
>reset the inode numbering?

Incremental dumps rely on the fact that files and directories don't
change their inode numbers during their lifetime.  This is so that they
can handle deletions correctly from incrementals.  E.g.:

mount /dev/whatever /a
mkdir /a/b
dump 0uf /level0 /a
rmdir /a/b
dump 1uf /level1 /a
umount /a
newfs /dev/whatever
mount /dev/whatever /a
cd /a
restore rf /level0
restore rf /level1

After this /a/b will not be present.  But you need to take a new level 0
dump so that later incrementals will do the right thing.

-- 
Christoph Badura					www.netbsd.org

	Anything that can be done in O(N) can be done in O(N^2).
	-- Ralf Schuettau (after looking at a particular piece of code)