Re: [rdiff-backup-users] Rdiff backup partial transfers

[Mark Ridley]

Hi Andrew,

Thanks for the reply.

Occasionally a new large file does appear that one of my users has created and I only have a 8 hours over an ADSL line to get that new file sent + other smaller file changes sent.

But also if a large file changes by quite a lot, the differences will also be quite large and will not get sent in 1 night so it would be great like rsync —partial if there was a way to restart the transfer next time from where it left off with that file.

Also, if a large file is interrupted, does all the changes to the other files that have been sent that night get rolled back? I.e is the whole lot an atomic transfer?

Thanks,

Mark


On 09/02/2009 14:44, "Andrew Ferguson" <address@hidden> wrote:

> On Feb 9, 2009, at 7:22 AM, Mark Ridley wrote:
>
> > Hi,
> >
> > The way rdiff-backup works is that if a file is being transfered and
> > the connection is lost, rdiff-backup rolls back the transfer of that
> > file.
> >
> > If the file is large - for example a database - it can take several
> > hours to completley transfer a file. If the connection gets dropped,
> > the transfer starts again from the beginning.
> >
> > Is there a way to do partial file transfers, so that the next time I
> > invoke rdiff-backup, it carries on where it left off.
> >
> > rsync works this way with the --partial option.
> >
> > is it possible to do this with rdiff-backup?
> >
> > if not, is this somthing that can be added with a code change - or
> > is there a fundamental reason why it cannot work this way?
>
>
> Mark,
>
> After the file is transferred once, it does not take very long for
> rdiff-backup to use the rsync algorithm to calculate any changes. As
> long as you are using rdiff-backup in the default configuration (with
> rdiff-backup installed on client & server and running over SSH),
> you'll be fine.
>
> That leaves us with the problem of the first transfer. If you are
> worried about the connection dropping during the first transfer, you
> can use rsync to copy the data to the server, then turn it into an
> rdiff-backup repository with the use of the --force option. I believe
> the Wiki describes this in greater detail.
>
> $ rsync foo rsync://address@hidden/backup
> $ rdiff-backup --force foo address@hidden::/backup
>
> then drop the --force option from subsequent rdiff-backup operations.
>
>
> If you think you'll be creating large files regularly after the first
> backup, then perhaps other users have some creative suggestions. I
> haven't thought about adding this feature (--partial) to rdiff-backup,
> so I don't know how feasible or not it is.
>
>
> Andrew