On Mon, May 17, 2004 at 11:37:27PM -0600, Bret McKee wrote:
> Greetings:
> 
> Inspired by what I think that is telling me, I would like to do is
> something like this:
> 
> - Run compressed backups to the holding disk
> - while the holding disk has > 1 tape worth of data (40 GB), write a
> tape.
> 
> It isn't clear to me how to set it up to get this behavior.  It seems
> like I might be able to get it by having two configurations, but then
> it isn't clear of the tape tracking database features will work like I
> would like (i.e. if one configuration writes to the disk only, and
> another sometimes pushes the disk to a tape, will the database know
> which tape the files from the disk only configuration are on?)


If you can put up with nightly error messages, you could run your
daily backups with amanda.conf containing a fake tape device name
(not /dev/null).  Couple of ways to handle this:

 - Have your cron job execute a script instead of just amdump
   which renames /dev/nst0 (or whatever) to /dev/_nst0, then
   calls amdump, and when done restores the device name.

 - Have amanda.conf edited each night by the cron script

 - Have two otherwise identical configs (use same index/log/
   ... dirs).  One with a fake device name will be used for
   backups.  The other with the correct name will be used
   for flushing and restoring.

At a daily time of your choosing (when amdump is not running),
run a cronjob that du's the holding disk and if sufficiently
large, runs amflush.

-- 
Jon H. LaBadie                  jon AT jgcomp DOT com
 JG Computing
 4455 Province Line Road        (609) 252-0159
 Princeton, NJ  08540-4322      (609) 683-7220 (fax)