[Bacula-users] Pool chaining / pool setup

[Uwe S. <uwe...@ni...>]

Hi folks,

in our current bacula setup (5.2.6 on CentOS 6), we run a load of full
backups starting on Friday evening which all go to the same online
disk file. The disk volume is currently 2,9T in size, so as a
consequence pruning / recycling the volume takes about six hours which
is a bit long (meaning that full backups won't start before midnight
or so even if we schedule them around 2000 Friday).

I'd like to resolve this issue by having each backup job go to its own
disk volume (maximum volume jobs = 1), using some decent naming
convention. 

I'll have to store two full backups and about a dozen incrementals for
each client, so I was wondering what the best pool definition would be
in this case.

Should I use pool chaining (I remember someone having problems with
chaining a while back) where a volume would be "migrated" to the
second pool once the full backup starts, or should I use a single pool
for all full backups? 

If the latter, how can I limit the number of volumes / disk space used
while at the same time making certain I have two full backups for each
client? 

Ideally, it'd like to limit the number of volumes on a "per job"
basis, but I don't think that's possible with bacula at the moment or
am I missing something? 

Thanks in advance for your time & suggestions, 

Uwe 

PS: Some bweb stats: 
Total clients:	128	Total bytes stored:	74.44 TB
Total files:	78168617	Database size:	112.55 GB