I'm not a bacula expert, but have had 30+ years in the industry doing backups. I'm a little concerned about what you are planning. As I understand it you are going to be keeping just one copy of each machine on a disk attached to the server. This will help if you loose the running disks (though it is hardly backup in depth), but what happens if you loose the server due to fire, flood, electrical problems, theft or even plain old dropping it? In general you should aim for multiple backup copies; are you willing to bet the company's future on one untried copy? You should ensure that the backup copies are held preferably off site, failing that in a separate building, or else in a secure fireproof strongbox.
Start by assuming you come into work one day to find the building burnt out and collapsed. Now work out how to rebuild your system on new kit on another site and you'll find that you define your backup needs.
Regards, Martin
On 12/10/16 12:54, Alessandro Baggi wrote:
Hi list, I'm building a backup server for 3 hosts (1 workstation, 2 server). I will use bacula to perform backups. The backup is performed on disks (2 x 3TB on mdraid mirror) and for each hosts I've created a logical volume with related size.
This 3 hosts have different data size with different disk change rate. Each host must have a limited sized resource and a reserved space. If a server needs more space to perform backup, It must be enabled and provisioned.
My first solution was put each host pools on different logical volumes, like:
host1 -> lv1 host2 -> lv2 host3 -> lv3
and store pools/volumes on specified storage daemon that uses a specified device for each different hosts.
host1 -> storage1 -> device_lv1 host2 -> storage2 -> device_lv2 host3 -> storage3 -> device_lv3
Unfortunately, I can't define on bacula-sd.conf multiple storage definition but only multiple devices. To use different storage I must run 3 bacula-sd on same host (I can?), run a bacula-sd on a vm/host. Ah, I must use only one physical server.
With one single machine and the current state I can't use multiple storage daemons.
There are other ways to store host volumes on different devices?
My second solution was, use only one storage daemon (on the same host) with a single device LVM over mdraid, create pool for each hosts and limit the size for each volumes on related pool.
Suggestions?
Thanks in advance.
CentOS mailing list CentOS@centos.org https://lists.centos.org/mailman/listinfo/centos