On Tue, 2008-07-29 at 17:19 +0200, Gabriel CORRE wrote: > I'm trying to do a differential backup. Each commands/scripts are launch on > backup server side. > > - rsync-ref: 1 script perform a complete backup to a directory "ref". > something like that: > rsync --archives \ > station:/path1/ /servbasepath/ref/servpath1/ > > - rsync-diff: 1 script perform a differential backup to a directory "diff". > something like that: > rsync --archives \ > --compare-dest=/servpath1/ \ > --delete \ > station:/path1/ /servbasepath/diff/servpath1/ > > > See the following scenario: > > on serv: > - rsync-ref is done > Result: large number of files are synchronized. > on station: > - the user create a new file "test" > on serv: > - rsync-diff is done > Result: the new file "test" was synchronized in diff directory. > - do rsync-ref > Result: the new file "test" was synchronized in ref directory. > - do rsync-diff > Result: nothing done, all file are synchronized. > That is my problem: > the file "test" in diff directory may/must be deleted, > because it is already on the ref directory. > > > I think --delete (or some other --delete-* options too) should remove > the file because it is already found in ref directory. > > But Is it really a bug ? > or may I request a --delete-if-found-in-compare-dir ?
Rsync is behaving as expected. The man page description of --link-dest
says, "rsync treats existing files as definitive (so it never looks in
the link-dest dirs when a destination file already exists)"; it would be
good to add that the same is true of --compare-dest.
You are welcome to request an option to make rsync delete destination
files that are matched in --compare-dest dirs. I would actually suggest
a more general option called --recheck-basis-dirs that also replaces a
destination file matched in a --link-dest dir with a hard link of the
basis file. (I don't think the analogous behavior for --copy-dest would
be useful.)
> For workaround I run a script that find duplicated copies of files on
> the both directories. And run the rsync-diff after.
> But I work on more than 6 millions of files, the rsync and find-scrtipt took
> a very lot of time.
> Rsync already compare the both dir with --compare-dest then
> it will probably easily append duplicated files to the delete queue.
Another approach would be to run rsync-diff to a new destination each
time, but I imagine you want to reuse the data in the old destination
instead of retransferring it all over the network. The command
rsync --compare-dest=/servbasepath/ref/servpath1/ \
--copy-dest=/servbasepath/diff/servpath1.old/ \
station:/path1/ /servbaseepath/diff/servpath1/
would do that, except that rsync currently doesn't support mixing
different types of --*-dest options on the same run; see:
http://lists.samba.org/archive/rsync/2007-February/017210.html
Matt
signature.asc
Description: This is a digitally signed message part
-- Please use reply-all for most replies to avoid omitting the mailing list. To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
