Re: [Duplicity-talk] Multiple backends (for instance local/network + S3)

[Simon Ruderich]

On Mon, Mar 15, 2010 at 09:49:49AM +0100, Olivier Berger wrote:
> I wonder if it would then be possible to restore from S3... probably...
> but this seems quite obvious, yes... maybe having to fiddle with the
> archive dir or archive name somehow...

As the file structure is the same (I hope so, haven't tested it
with S3) you just change the duplicity restore command to point
to S3 and it should work.

> But in general, my idea would be that it is integrated into duplicity,
> and is run from the source side (not the "local" target's side) so that
> in case of unreachability of one of the backends, duplicity would manage
> this automatically, for instance. Also there wouldn't be problems like
> sync to S3 started while a backup is still in progress on duplicity
> side, etc.

I'm not a Duplicity developer but I don't like the idea of adding
this to Duplicity, IMHO it doesn't add much but is not so easy to
implement right (see below).

I don't know how S3 works, but if you can use something like
rsync the sync is easy and fast and as Duplicity only adds files
it should be possible to do even while Duplicity is running.

> A use case would be that whenever at "home" you backup to both
> "local" (LAN) and cloud (S3), and whenever on the go, you still backup
> to the cloud.
>
> If the sync is done on the local dest host (in case of LAN backups),
> then S3 would be updated only when you've cehcked-in back home firs.

Yes, that's true. But if you have different backup versions (one
in local LAN and on on S3) you must do a complete backup using
duplicity anyway, so this can't be really done in one duplicity
run.

Why not just run Duplicity two times, one for local LAN and one
time for S3?

> Does it make sense ?
> Thanks in advance.

Simon
-- 
+ privacy is necessary
+ using gnupg http://gnupg.org
+ public key id: 0x92FEFDB7E44C32F9

pgpggUEjo0ifx.pgp
Description: PGP signature