[Borgbackup] Repo sync efficiency

Dmitry Astapov dastapov at gmail.com
Tue Oct 31 09:12:43 EDT 2017


So if I have append-only repo and created archive archive-A with ./file/A
in it that occupied more than one segment (lets say those were segments 100
and 101), and then created more archives where that file in absent, and
then 'borg prune'd repo::A, the following will happen:

- Segment 100 (which contained data from ./file/A and nothing else) will be
no longer needed and will be physically removed from disk
- Segment 101 (which contained data from ./file/A + some other files as
well) will be kept around and will not be compacted/rewritten to get rid of
the chunk that corresponds to ./file/A

Or am I wrong for segment 100 will be kept around even after archive-A is
pruned?

On Tue, Oct 31, 2017 at 12:03 AM, Thomas Waldmann <tw at waldmann-edv.de>
wrote:

> > Can I clarify something?
> >
> > append-only more repo would still delete unused segment files once the
> > last archive that references then is pruned, would it not?
>
> A append-only mode repo looks like doing everything normally from the
> outside, so you can create and delete repos as usual.
>
> borg always appends new stuff (PUT, DELETE, COMMIT tags) at the end.
>
> The difference is that in append-only mode, there is no compact_segments
> that removes deleted / superceded PUT entries and rewrites non-compact
> segments into compact ones.
>
> So, the previous repo state is always conserved also when you only look
> at the first N segment files.
>
>
> --
> GPG ID: 9F88FB52FAF7B393
> GPG FP: 6D5B EF9A DD20 7580 5747 B70F 9F88 FB52 FAF7 B393
>
> _______________________________________________
> Borgbackup mailing list
> Borgbackup at python.org
> https://mail.python.org/mailman/listinfo/borgbackup
>



-- 
D. Astapov
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/borgbackup/attachments/20171031/b5a7d25a/attachment.html>


More information about the Borgbackup mailing list