[CentOS] C7 Backup problem

Tue Aug 9 17:15:12 UTC 2016
Valeri Galtsev <galtsev at kicp.uchicago.edu>

On Tue, August 9, 2016 11:51 am, Gordon Messmer wrote:
> On 08/08/2016 11:11 PM, Alessandro Baggi wrote:
>> |    78 | backup-fd | 2016-08-08 09:52:57 | B    | F     |  191,024 |
>> 7,978,295,072 | T         |
>>
>> |    79 | backup-fd | 2016-08-08 12:10:31 | B    | F     | 0 |
>>          0 | A         |
>>
>> |    80 | backup-fd | 2016-08-09 03:05:03 | B    | I     | 191,024 |
>> 7,978,337,100 | T
>>
>>
>> Job with id 79 was aborted due to disk failure.
>>
>> Why, if I have a job full performed on sunday, on monday I got an incr
>> job with the same size?
>
>
> As far as I know, incremental backups include all changes since the last
> full backup.  You do have a full backup on 8/8 at 9:52, but that isn't
> the *last* full backup.  The last full is 8/8 at 12:10, and it doesn't
> include anything, so all files are changes since the last full backup.

Well, actually incremental is only the difference between current state
and last backup (whichever it was: incremental, differential, or full).

In other words, to restore the whole thing to today's date you do:

1. restore everything from latest full backup

2. restore everything from latest differential backup, which is difference
between latest full backup state and machine state on the day of that
differential backup was performed (if you have more than one diff backup
since full, you choose latest)

3. restore everything from all incremental backups performed after last
differential in chronological order


Of course, these are the definitions of full, differential, and
incremental that bacula (or bareos) uses. When you are restoring some file
or directory on particular day/time bacula (or bareos) does all necessary
lookups in database to track
full-->differential-->incremental-->...-->incremental history for
particular object (file, symlink, directory,...) and only restores
relevant copy, the one after which object didn't change till requested
day/time.

I know the terminology they use is a bit confusing, but I hope the scheme
above helps un-confuse it.

Valeri

>
> Naturally, I could be wrong, but that's what it looks like from here.
>
> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> https://lists.centos.org/mailman/listinfo/centos
>


++++++++++++++++++++++++++++++++++++++++
Valeri Galtsev
Sr System Administrator
Department of Astronomy and Astrophysics
Kavli Institute for Cosmological Physics
University of Chicago
Phone: 773-702-4247
++++++++++++++++++++++++++++++++++++++++