On Thu, Dec 9, 2010 at 3:05 PM, Kevin Fries
<kfries6@gmail.com> wrote:
On 12/09/2010 02:49 PM, betty wrote:
thx, steve,alan, joseph and anyone i forgot !
good pick up ; i AM doing backups with names of the month, so it IS backing up everything each time since it is a different name. i understand that but was doing it because i was so paranoid that it would miss something ... so i have two months backups on the ext hdd - oct and august. usually when i do the next backup, i put in the new month and delete the oldest one. but once again, it is backing up everything of course.
so, should i try to rename the october file then just always use that for backups. ? or should i just call everything october from now on, hehehe. fine with me, easy to remember.
Create a directory for each machine to be backed up
Create a directory for each backup to be performed (one per day/week/etc)
Copy all links (hard links) from previous backup
Read a file from the target over ssh/rsync
Is this file already in the backup?
Yes - Has It Changed?
Yes - Add file to pool, overwrite old hard link to point to new file
No - Leave link alone
No - Leave link alone
Read next file and continue
This causes the system to do a incremental backup every day, but it acts like a full backup. Files are removed from the pool once it is A) deleted & B) no longer referenced from an old backup. How long it keeps backups and when it creates a new file regardless (i.e. full backup) is configurable. This program is especially useful for backing up multiple machines (as it is smart enough to recognize same files from multiple machines), and I know this is not your case... But I thought maybe the workflow could be useful for you to look at in designing your own backup policy.
Kevin Fries