|
Hi,
Has anyone thought of making rsync
object-oriented? For instance, I just want to slightly modify a few of its
algorithms for somewhat specific backup purposes (rdiff type logic for
incrementals and pooling files for many source machines with hardlinks -
similar, but not exactly the same as, BackupPC). I'm currently using
wrapper programs to keep metadata in a BerkeleyDB database but I would rather
just inherit rsync's methods and over-ride them a bit to check metadata and not
the file itself. I figure I might want to keep the whole-file checksums as
well instead of using md5 on the backup machine end. There are other
things I would modify as well. It seems to me that this would greatly
increase the uses for rsync.
I'd even like to create a linux virtual filesystem
on the backup machine to seemlessly work with this logic. It would
have to have multiple inode structures for whichever machine wants to get or put
the file for instance ... somewhat similar to hardlink logic but with a
"source machine" type field added. For instance, if machine A wants file
x, a stat() would return a stat appropriate for machine A (uid, gid, mtime, ...
etc). Machine B would get a different stat(). Has anyone thought
along these lines?
BTW, I just joined this board and don't know how to
search through it. Is the only way to do this to download each month of
posts and manually search each file?
--Blake
[EMAIL PROTECTED] |
-- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
