I understand that Rsync does a full backup the first time it is run and then for each following occurence, only the changed files are copied and the deleted files removed off the target (if the option is checked).
However, what does the option to do a full backup every x weeks achieve? I have this checked but was thinking, it won't delete the files on the target like a CIFS backup for example.
When a full backup is performed the backup destination would be wiped and then all the data transferred. I would recommend configuring the full backup to be run first time only when using the Rsync option especially if you are backing up to a remote location via the internet.
The NAS keeps a record of when the last full backup was run. So it shouldn't do a full backup next time if it's set to do a full backup first time only.
As I use rsync now to mirror to NAS devices, should I ever have them do a full backup or is the fact that it did it once enough. Is there any benefit in doing a full backup? I have the delete files on target option selected.
RSYNC skips over files that match the size and date. So if the backup device develops a bad sector within one of those files the failure will not be detected until you try to read that file off the backup. I had something similar to that happen on a USB drive (with Windows) using incremental backup.
So there is some value in doing a full backup from time to time. Or alternatively read all the files on the backup device, just to be sure that you can.
I do not think it is necessary to do an entire backup every couple of weeks. When running rsync in a shell you can add the parameter
-c, --checksum
which means "skip based on checksum, not mod-time & size" according to rsync documentation.
That should be way faster than copying everything.
I wrote a script that simplifies calling rsync in a shell or by cron. All parameters are set in a backup job file which is fed to the script. You may download it on my blog: http://goo.gl/d3VKd
I would be happy if you could leave some feedback there after you have tested it. :wink:
--checksum would also do the trick, and would reduce bandwidth. It would be pretty slow though (compared with incremental mod-time+size), since all the files in the share need to be read (and checksum computed) on both the client and the server. So this also probably doesn't want to be done on every backup.
