Am 11.05.2015 um 20:49 schrieb Alessandro Baggi alessandro.baggi@gmail.com:
Hi there, In my last request I have asked info about backuppc and other backup solutions. After some test I have choosen bacula. Many people said it's too complex and try to make it works is a challenge. I've tested backuppc and I don't like it for a stupid reason... I must install on centos external software, configure a web server with cgi with suid enabled (apache does not have mod_perl) I must exchange ssh key or configure smb shares. On backuppc site is reported that there is no database because it is another point of failure. But to make backup pc working I need different softwares to make it work. With bacula I need pgsql (installed with 3 step) install bacula with yum and configure bacula. Bacula is amazing in it's configurantion...you must spend time for studying it but it is solid and powerfull. This is my choice and this not mean that backuppc is bad.
Then after this...I'm new to bacula.
On different sites, on bacula docs and on sample files, there is specified a job related to database (catalog) backup. The catalog is used to restore a backup.
What do you think about backupping catalog on the same backup server?
The scenario is a baremetal recovery for example. For this the "volumes" (e.g .tapes, hdd) should be at hand, and to help the restore process also the "catalog" ...
If the backup server dies, how I can restore catalog(s)?
the catalog is just a db dump ... so restoring a catalog implies setting a db service up. This helps to browse the meta data and to find the right volume for restoring data.
If server dies, I can restore without reinstall bacula?
its possible to "manually" dump your files from your volumes but that is less flexible. We use a process for baremetal restores where we boot a live-system with an "bacula" service enabled and import then the newest catalog.
What solution do you use?
Check your use cases. In any case, the catalog helps to restore files. Therefore after "all" bacula jobs we have a catalog job to make sure that we have a current catalog on the volumes, additionally we store the catalog dump outside of the volumes (RunAfterJob).
-- LF