Borg: Difference between "Deduplicated size" and real size of the repository

[Thuffir]

borg 1.1.4
Python 2.7.9
Linux xxx #1 SMP Debian 4.9.30-2A~4.2.0.201801290155 (2018-01-29) x86_64 GNU/Linux

borg info shows me:

                       Original size      Compressed size    Deduplicated size
All archives:                8.51 TB              5.36 TB            337.17 GB

but when I do a df -hs on the repository I get only 315G.

Is my assumption that the Deduplicated size of the whole repo is the actual size wrong?

[stevesbrain]

Presuming you've done a data check, you'd be likely looking at possible filesystem/block size differences (at a guess)

[Thuffir]

Not sure what you mean, but du -hs --apparent-size shows me also 315G.

[stevesbrain]

In that case, not sure sorry :) I'm a user of borg, rather than someone who's delved into the code. I'll let someone more knowledgeable come up with a reasonable explanation!

[infectormp]

@Thuffir did you read https://borgbackup.readthedocs.io/en/stable/usage/info.html#description ?

[Thuffir]

Yes, and it says:
All archives / deduplicated size = amount of data stored in the repo = all chunks in the repository.
And that is why i though that it must be the "real" size of the repository.

[ThomasWaldmann]

>>> 337.17 * 1000**3 / 1024**3
314.0140324831009

borg's unit is GB (1000^3), linux tools use GiB (1024^3, advertised wrongly as GB).

[Thuffir]

Hehe. Thanks for clarification. Makes sense :)