Hi George,

Just to summarize what has already been said, if I'm understanding correctly 
you have three options:

1. For a client with a temporary file system that should never be backed up:
Create a directive to skip or null the temporary file system, then back up the 
client with Save set: All using that directive.

2. For a client where some file systems are to be backed up to one pool and 
some to another, different file systems at different times or on different 
schedules, etc., but you also want to be able to back up new file systems when 
they're created, even if you're not notified about them:
Create client definitions with enumerated save sets using the respective pools, 
schedules or groups. Then create a directive to skip or null all the enumerated 
save sets across all the client definitions. Finally, set up an additional 
client with Save set: All using that directive. Normally that last client 
definition won't back anything but if a new file system is added it will be 
backed up that way. You could use a pool for that last backup that doesn't have 
any media assigned to it; that way when a new file system is created, it won't 
back up but you'll get a "Waiting for volume in pool ..." message that will 
have to be attended to. At that point you kill the backup, add the new FS to 
the save set list in the appropriate client definition and to the directive and 
you're back in business.

3. You could definitely write a script to do dynamic save set selection. 
Probably the best way to get a list of file systems for a given client is to do 
a probe (savegrp -p -c <client> -G <group>). Then the script would morph that 
list -- grepping out what you don't want to back up -- into a file like this:
. type:nsr client;name:<client>
update save set: <save set list>;
<blank line>
Then update the client definition by nsradmin -i <file name>
If there are multiple client definitions then the first line becomes
. type:nsr client;name:<client>;group:<group>

HTH,

Conrad Macina