Re: [Duplicity-talk] New proposed feature removing incrementals of old full backups

[Olivier Berger]

Hi.

FYI, Kenneth has been kind to merge in my proposed implementation for
that new feature into :
https://code.launchpad.net/~duplicity-team/duplicity/0.6-series

Any testers much welcome ;)

Best regards, and thanks for the #666 commit (yeah, won't harm your
backups, I swear ;)

Le samedi 10 avril 2010 à 11:00 +0200, Olivier Berger a écrit :
> Hi.
> 
> (It seems no one really saw my previous message so restarting with a new
> thread)
> 
> I've tried and implement a new duplicity command that would allow to
> remove incremental backups from old backup chains [0].
> 
> It seems it's working from the new tests I've made (previous problems
> were caused by file access rights I think).
> 
> May I ask for some code review of the bzr branch for my proposed
> implementation at [1] ?
> 
> The goal of this new option is to allow some backup "retention" policy
> like where only increments of the last month would be kept, and for
> older times, only monthly full backups (next step will be to add that to
> backupninja's duplicity handler, I think).
> 
> Many thanks in advance.
> 
> Best regards,
> 
> [0] : https://bugs.launchpad.net/duplicity/+bug/542482
> [1] : https://code.launchpad.net/~olivierberger/+junk/dupl-542482
> 
> -------- Message transféré --------
> De: Olivier Berger <address@hidden>
> Reply-to: Discussion of the backup program duplicity
> <address@hidden>
> À: Discussion of the backup program duplicity
> <address@hidden>
> Sujet: Proposed implementation - review request Re: [Duplicity-talk]
> Removing incrementals of old full backups
> Date: Mon, 29 Mar 2010 08:58:06 +0200
> 
> Hi.
> 
> Le samedi 20 mars 2010 à 11:53 +0100, address@hidden a écrit :
> 
> > > 
> > >> 3) there is currently no way to cherry pick delete some backups. If you
> > >> really want to you can have a look in your backup repository and
> > >> manually/scripted delete the backups you do not need anymore. The file
> > >> names are pretty descriptive.
> > >>
> > > 
> > > Hence the need for some kind of delete-old-incrementals command IMHO.
> > 
> > as always, go ahead and implement it or (as you did) file a feature request
> 
> -> https://bugs.launchpad.net/duplicity/+bug/542482
> 
> I've tried and implement it, then.
> 
> You'll find in lp:~olivierberger/+junk/dupl-542482 a proposed
> implementation, which consists as a new command
> "remove-all-inc-of-but-n-full" which is a variant of
> remove-all-but-n-full.
> Only at the moment of the deletion of the set, will it check whether it
> is actually incremental (and delete) or full (and will then keep it).
> 
> It should work... however, I can't seem to see the files deleted really
> from a local backup backend :-/
> 
> I wonder if it is me or if the code of remove-all-but-n-full wouldn't
> work either... :-/ something as the delete() function of sets doesn't
> seem to actually unlink anything on the disk. Haven't had time to debug
> any more, but any advise would be much welcome.
> 
> Hope this helps.
> 
> Best regards,
> 

-- 
Olivier BERGER <address@hidden>
http://www-public.it-sudparis.eu/~berger_o/ - OpenPGP-Id: 2048R/5819D7E8
Ingénieur Recherche - Dept INF
Institut TELECOM, SudParis (http://www.it-sudparis.eu/), Evry (France)