[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Duplicity-talk] checksum rsync option (duplicity 0.6.21)
From: |
Laurent Lavaud |
Subject: |
Re: [Duplicity-talk] checksum rsync option (duplicity 0.6.21) |
Date: |
Mon, 22 Jul 2013 17:11:43 +0200 (CEST) |
ok thanks for the explanations, i can remove my useless rsync option !
and i have found the guilty option in my tag tool (puddletag), "preserve file
modification time", it is weird to add an option like this...
--
Laurent Lavaud
Administrateur Systèmes et Réseaux
----- Mail original -----
> On 22.07.2013 14:23, Laurent Lavaud wrote:
> > Hello,
> >
> > I would like to use the rsync checksum method to detect files
> > modification because the "classic" method do not detect some
> > changes, like mp3 tag modification (mtime not modified)
> >
> > So i have tried to add --rsync-options="--checksum" to my duplicity
> > command line:
> >
> > /usr/bin/duplicity -v9 --encrypt-key=xxx --sign-key=xxx --use-agent
> > --allow-source-mismatch --rsync-options="--checksum" test
> > ssh://address@hidden/backup/test
> >
> > But duplicity doesnt seem to use it because it not detect the file
> > modification...
> >
>
>
> --rsync-options only applies to the rsync backend. the internal
> librsync routine is not configurable. actually duplicity itself
> decides if the file changed and invokes librsync internally then.
>
> duplicity compares wrt. time only mtime
>
> http://bazaar.launchpad.net/~duplicity-team/duplicity/0.6-series/view/head:/duplicity/path.py#L306
> for performance reasons.
>
> only verify can compare data as well. this is a design decision.
> duplicity backup data is usually stored remotely. comparing data
> during backup would need to restore the former version of the file
> locally and therefor download all volumes necessary resulting in
> lot's of traffic.
>
> what i describe above is merely the status quo. i am aware that in
> theory a checksum could be saved remotely but i cannot find anything
> like that in duplicity's source. the tarfile info used as metadata
> store also does not seem to support an entry like that.
>
> Ken, Mike: is the above correct?
>
> Laurent: wrt. your issue above. the behaviour of your tagging
> software or filesystem is the "real" problem here. usually
> filesystems modify mtime automatically on every write access. if
> not, something weird is going on and you should research what's
> going on there. e.g. we had users on the list where mounted remote
> filesystems did not update file stat data correctly.
>
> good luck ..ede/duply.net
>
>
>
>
> _______________________________________________
> Duplicity-talk mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/duplicity-talk
>