[Duplicity-talk] Questions about Duplicity

[John Goerzen]

Hi folks,

I'm looking into whether to use duplicity for my backups.  I have a
few questions about duplicity after having read the manpage.

1) Why does Duplicity split the tar files into 5MB chunks by default?

2) What are the exact semantics regarding storing of hard links?  I
   gather rdiffdir can't do this, but I'm wondering what exactly that
   means.  Does it mean that it just ignores the entire concept of
   hard links, or that it can handle hard links within the same
   directory but not those without?  Is there any way around this, or
   is this a fundamental problem?

3) It appears that Duplicity supports only full and incremental backup
   levels.  Are there any plans to support, say, differential backups,
   or even backups of arbitrary levels?  This could make some sense
   for minimizing the number of deltas that must be read, while also
   keeping storage requirements down.

4) Say that my last full backup was a month ago, and I've made
   incrementals every day since then.  Duplicity will then have to
   read through 30 deltas.  Does it do this via pipes (say, opening 30
   connections to the remote and streaming it down to rdiff
   processes), or does it require lots of temporary space to stage
   this?  If it requires temporary space, how does the amount of space
   relate to the size of the backups?

5) Is there any support for creating a full backup checkpoint by
   playing the history forward, rather than uploading a full backup?
   This could be useful if there is shell access to a remote machine,
   where checkpoints could be created via a local duplicity
   installation rather than sending gigabytes of data over the net.

Thanks,

-- John