On Tue, Aug 12, 2003 at 08:00:40AM -0400, Dalton, John L MONMOUTH ITS Multimax 
wrote:
> Is this saying I have most of my DLT tape left
> and I cannot append to the tape because it is an
> active tape?

Correct.  Amanda refuses to append to a "live" tape on a
subsequent run.

The main reason, I think, is the risk of accidentally clobbering
the backups already on the tape.  The "append" could turn into an
"overwrite" (e.g. due to an off-by-one error in Amanda's
fast-forward code, or in the tape driver, or due to a positioning
error in the tape drive's firmware, or a problem in the
tape-transport hardware) or the drive could eat the tape
completely.  I'm sure there are other failure modes as well, any
one of which would trash those live backups.

John Jackson, who used to be a leading Amanda maintainer, often
referred to tapes as "evil" devices, and not to be gambled with.
To paraphrase his formula for writing a tape safely:
  - start writing at BOT
  - exactly one forward pass
  - *never* back up
  - *never* fast-forward
  - don't trust that the data's been written to the media until
    you've *successfully* closed the tape device -- it could be
    buffered in the kernel or in the drive, and *only* a close()
    call can be trusted to both flush the buffer and return a
    meaningful status (though of course that doesn't absolve you
    from checking the write()s' statuses too!)
  - after all of that, cross your fingers and hope like &*@(
    nothing's gone wrogn

--

|  | /\
|-_|/  >   Eric Siegerman, Toronto, Ont.        erics AT telepres DOT com
|  |  /
When I came back around from the dark side, there in front of me would
be the landing area where the crew was, and the Earth, all in the view
of my window. I couldn't help but think that there in front of me was
all of humanity, except me.
        - Michael Collins, Apollo 11 Command Module Pilot