Re: [Duplicity-talk] duplicity and partial rsync transfers

[Eliot Moss]

On 8/29/2011 11:43 PM, Martin Pool wrote:
> I think the short answer is that it's probably easier to just leave
> the volume size fairly small.  With 25MB volumes and if that's
> something like 2500 volumes (assuming a bit of compression) and that's
> not too much to have in a single directory or to retrieve as a single
> file listing.
>
> If duplicity is interrupted and restarted and if the files affected
> have been touched at all it's likely not to have anything in common
> between the volumes in their compressed form, and so rsync might not
> be able to do an incremental transfer.
>
> That said passing --partial seems probably worth setting except then
> duplicity will hit more situations where the volume is apparently
> present but actually truncated; in theory it could cope with that but
> I might expect some fall-out bugs.

Thank you, Martin.  I think I see what you mean in the case that
the sending side stops and has to be re-run. Still, if nothing
has changed for the volume in question, the compressed encrypted
contents of the last volume should be the same. So here --partial
might make some sense.

I was also concerned about the case where the sender is still
running, but the network connection was interrupted so that it
starts rsync again, but on the same volume file.  This strikes
me as a case where --partial makes a lot of sense.

In either case, if the volume to be sent changes, --partial
does no harm, but probably no good either, though since
encryption is sequential (from beginning to end) I would
think that the encrypted volumes might still often have a
common prefix, so --partial would help if the number of
changes is small.

Regards -- Eliot