[rescue] Mounting and Dumping

[Sheldon T. Hall]

 Mike Meredith says ...
> On Wed, 14 Jan 2004 16:42:42 -0500, Sheldon T. Hall wrote:
> >
> > Of course, what's _really_ bad is that ufsdump doesn't tell you it's
> > writing crap; you only find that out when you try to restore from your
> > backup.
>
> True enough. Mind you some of the others can be just as bad ... see
> http://berdmann.dyndns.org/zwicky/testdump.doc.html. One of my
> coworkers says that the one thing he doesn't like about Unix is that
> there's multiple ways of doing backups and all of them seem to have
> problems (and likes to point out that VMS has just one that works).

<kinky friedman>
It always fails to amaze me ...
</kinky friedman>
... that the folks who buy [unix|windows|whatever] put up with this.  I
mean, how hard can it be to back up the friggin' files in a sane and
reliable manner?  In 1983 I wrote a backup program for CP/M, in BASIC, that
would back up any size of hard drive, to any number of floppy disks, and
arrange the files to fill the floppies as full as possible.  If I could do
that, why can't the guys at [Sun|Microsoft|wherever] do something better
twenty years later?

OK, I'll stop now.

> > >  (Read all the way through all the responses to that one)
> >
> > No consistent message there ... some say it's OK, some say it's not.
>
> That's the problem. I suspect it's pretty rare that there's a problem,
> but then again I suspect that there's plenty of people having problems
> with ufsdump that don't notice ... because they don't test their
> backups.

Heh, I'll bet almost no one does that, actually.

> Stopping syslogd could be something that's overlooked in some
> of the problem cases.

Hmmm.  Yeah, but then where's ufsdump gonna write its log entries?

> > I find it interesting that SGI does things quite differently.  xfsdump
> > will_only_ dump mounted filesystems, according to the man page, and
> > the SGI GUI backup thingie uses cpio rather than xfsdump anyway.
>
> They aren't the only ones ... the AdvFS dump for Digital Unix also
> requires a mounted filesystem. Interestingly the IRIX dump for efs
> requires you to unmount the filesystem ... I guess SGI had enough
> complaints about this for fixing dump to be an item on the list of
> things to do for XFS.

I wonder if they got more complaints about this than Sun gets.

> But perhaps {I am] a little paranoid about backups. Having once had a
> 20-hour day trying to restore a system (and encountering bad tapes)
> with the Director of Finance having a nervous breakdown behind me
> has left a lasting impression :)

Heh, I'll bet.  Those guys can be remarkably uptight under those conditions.
He probably still has the flop-sweat stains in that suit.

> > Remember those sucky little Colorado
> > drives that ran off the floppy disk cable on a PC?
>
> Still got one around here somewhere, and a collection of the tapes
> sitting in my desk at work. Even better I've got a 9-track tape on my
> desk to scare PFY's with.

I wish I'd had the wit to save a deck of CRAM cards and a couple of miles of
paper tape from my first job in "DP."  My merely explaining real core memory
or a CRAM unit doesn't have the same impact as the PFY's seeing one in the
flesh ...

"... and this is one of the core memory cabinets.  We have seven of these,
each with 8,192 12-bit "slabs" of memory.  That means over ninety-eight
thousand individually wound bobbins connected to ..."

"JEEEZUUSSS, LOOK AT ALL THOSE WIRES!!!!!!!!"

"... and every one connected by hand.  Now, these four cabinets over here
..."

We had a great backup plan there.  We'd copy yesterday's data to today's
CRAM decks, and put yesterday's decks in the vault.  We'd re-use the decks
each week, except for the Friday deck, which we saved for a month, and the
end-of-month deck, which we saved for a year.  Then there were the
printouts....

-Shel