On Mon, Dec 30, 2002 at 01:12:22PM +0100, Georg C. F. Greve wrote:
> Hi all,
> 
> I use amanda in a "mostly on hard disk" setup. 
> 
   [[ snip ]]
> 
> The script first catalogues all files on disk, then looks for the
> obsolete ones (eg: a fresh level 1 dump makes all the preceding dumps
> with equal or higher levels obsolete) and deletes them. 
> 
> An additional variable can be used to determine the wanted duplicity,
> which is called "keep." Example: If keep is set to 2 and the last dump
> was level 3, the program will delete all dumps with levels of 4 and
> higher, while keeping the last and second-last level 3 dump.

If I understand your algorithm, it would defeat one of the
amanda capabilities I like and have used a number of times;
the ability to recover files "as of a particular date".

For example, not too long ago I badly mangled a boot-time
configuration file.  Did not realize any problem until a
reboot several weeks later.  As I keep 4 dumpcycles worth
of backups (tape, not hard disk, but the concept would be
the same), I just recovered the config file by picking a
date 3 weeks earlier, prior to my botch job.

Just food for thought.

-- 
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)