On Mon, May 03, 2004 at 10:01:47AM +0200, Paul Bijnens wrote:
> Kianusch Sayah Karadji wrote:
> 
> > [...] the plan was to not put in a
> > tape, and have amanda backup everything to the dumps Partition. After that
> > use amflush to move the data from the partition to tapes in several runs.
> 
> Why so complicated?
> Let it dump to one tape, and flush the rest in the morning to the other
> tape(s).

I do our all-level-0 configurations this way, but for a different
reason: it more fully packs the first tape.  Since taping doesn't
start until all of the DLEs have been dumped, the largestfit
algorithm has all the data it needs to fill the tape as full as
possible.  If instead we let amdump write the first tape,
largestfit would pick the optimum DLE <i>from among those
available at the moment the drive became free</i>, which would be
suboptimal for minimizing tape usage.

I'm not sure whether this applies to 2.4.3, since I don't know
what (hardwired) tape algorithm it used.

> runtapes 1
> maxdumpsize 90g

That's what I did, but in 2.4.3, I presume one could get the same
effect by increasing runtapes instead.  In this situation,
runtapes doesn't seem to have any direct effect on the actual
number of tapes used :-/

> reserve 0
> taperalgo  largestfit

That's what we have.

--

|  | /\
|-_|/  >   Eric Siegerman, Toronto, Ont.        erics AT telepres DOT com
|  |  /
It must be said that they would have sounded better if the singer
wouldn't throw his fellow band members to the ground and toss the
drum kit around during songs.
        - Patrick Lenneau