On Thu, Mar 31, 2005 at 09:06:39PM +0000, Mark Costlow wrote:
> 
> I've got a situation where I'm probably backing up too much data for the
> size of "tape" I'm using (I'm using the file driver, but with relatively
> small 40G vtapes).  I believe the core of my problem is that several of my
> DLEs are producing pretty large incrementals (a few 5G, some 10G,
> occasionally 15G).
> 
> I'm taking steps to remedy the situation, but it has led to some behavior
> that I don't understand and wanted to ask if anyone has a good explanation
> of it (or something I might be doing wrong).
> 
> The issue is that these large incremental dumps seem to "starve" the planner
> and prevent it from doing full dumps.  In general I understand the argument
> is that it's better to get that nightly incremental and postpone the full
> dump for a while.  But in my case, there are several DLEs whose full dumps
> have been overwritten and are not getting replaced.  It seems that a full
> dump for a DLE that doesn't have any full dumps in the db should take
> precedence over any incremental dump.  I would rather miss one day of
> the incremental of a big DLE than be skating out here on thin ice with
> no level-0 of another DLE.

Have your reports included any info about why level 0's of those DLE's
have not been done for a month and a half?  Might the reports been
telling you that the DLE is bigger than the size of the tape?  If that
is the case, you will never get another full dump.

You might also look at the dumporder param, setting some capital esses ("S")
first I think would cause the largest dumps to run first.

I'm not certain, but I think priority only affects incrementals, but that
might be another thing to read up on.

-- 
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)