There are probably many ways, one idea I just figured out (haven't tested -- no idea if it's good really) is that you could mount the first machine to some place and create a level 0 backup of it, then umount it and mount the other one to the same place and create a level 1 backup of it (this might work if the systems are mainly identical, with minor changes). But if the systems have only some common files and mainly do differ, it's probably not going to work.
Another way around would be to use find or similar tool to produce a complete file listing of the first machine, then do the same trick for the second machine and use some program (diff if you'd like to just see which files differ) to see the differencies. And if you like to go further, mount both machine's filesystems so they're accessible on one machine, create the file listings I mentioned earlier and use sed (or any other tool that suits) to take out duplicate lines, creating a third "unique" file out of this information -- then create a backup based on this list. The problem here is that the same files do get backed up if they sit in different directories, but either you do want to back them up too, or you can think of a way to filter out the lines from the list that describe the same file (remember to take into account dates, file sizes, last modification dates etc).
I'd probably mount the machines under /mnt/backupdata/ into separate directories, create lists of files under these two (2 machines) directories (filenames, sizes, last modification dates or whatever you think is important) and start merging these files, dropping duplicate files' lines out, and create a backup based on this list. Sounds a bit difficult and might have some drawbacks, but if you really think you can't take two identical files backed up, it's something like this you need to do.
I think any information that can't be installed straight out of a cd/other media, without any modifications, is important and should be backed up; if there happens to be some files backed up twice it's ok, since the point is that the information is there. If everything goes well till the end of the world, you never need the backups and it's all the same how many duplicate copies there are of some files, but if you do need the backups, it's just good if there is information (even if it's duplicated - or especially then).
|