First of all, thanks for all the responses.
I guess I didn't end up describing the problem very well. Some people got
confused with my description, so I'll describe it again, then give the
answer.
I have a 5GB 8MM Exabyte on host a and a 2.3GB Exabyte on host b.
My intention is to use the high-density drive on host a to do level0
dumps for any hosts that had filesystems requiring backups, and use
the low-density drive on host b to do level5 dumps Monday thru Friday. All
of these dumps would be executed via cron scripts.
The problem was that the level5 dumps were acting as if there had been no
level0 dumps executed recently. As many of you described, I expected the
/etc/dumpdates file on each host to reflect that a level0 dump had been
done on whatever filesystem, but the entries just weren't there. This
was when I started wondering if I had the whole concept wrong. This was
when I started thinking that maybe I have to have a common /etc/dumpdates.
But I couldn't see how it could be done (answer: it's not supposed to be
done!). And that's when I wrote.
After a few responses came in, I again asked myself: "Why is there no entry
for the level0 dumps? The answer: in dump and rdump, the "u" option must
explicitly be included in the options list. This is a simple little thing,
but it has a lot of impact. Now that the u is in the scripts, it's working
as one would expect. As long as all invocations of dump and rdump have the
u option, it looks like things will be okay.
This begs the question of why the default situation is not to update
/etc/dumpdates. From my perhaps-naive perspective, it seems counterintuitive
to require the system manager to ask for the dump record to be updated.
I would think that you should be required to tell dump/rdump NOT to
update dumpdates. Perhaps somebody else knows the enlightened reasoning
for why things are the way they are, but that can be saved for another
day.
Thanks to all the respondents below. I hope I didn't leave anyone out.
Harry Edmon <harry@atmos.washington.edu>
"Anthony A. Datri" <datri@concave.convex.com>
Steve Romig <romig@cis.ohio-state.edu>
lars@CMC.COM (Lars Poulsen)
macphed@dvinci.usask.ca (Ian MacPhedran)
dal@gcm.com (Dan Lorenzini)
johnb%edge@uunet.UU.NET (John Benjamins)
uunet!deltam!dm!mark (mark galbraith)
todd@flex.Eng.McMaster.CA (Todd Pfaff)
Tom Conroy <trc@NSD.3Com.COM>
Ray Brownrigg <Ray.Brownrigg@isor.vuw.ac.nz>
kwthomas@nsslsun.nssl.uoknor.edu (Kevin W. Thomas)
baumann@proton.llumc.edu (Michael Baumann)
Gregory Higgins <higgins@math.niu.edu>
doug@perry.berkeley.edu (Doug Neuhauser)
Jim Mattson <mattson@cs.ucsd.edu>
Kevin Elphinstone <kevine@vast.unsw.edu.au>
unruh@physics.ubc.ca (William Unruh [Unruh])
Postmaster <Piete.Brooks@cl.cam.ac.uk>
alastair@eucad.co.uk (Alastair Young)
Doug Acker <dla@se05.wg2.waii.com>
plott@lpi.liant.com (Peter Ott)
"Patrick M. Landry" <pml@cacs.usl.edu>
uunet!trdlnk!mike (Michael Sullivan)
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:36 CDT