[Borgbackup] Rebuilding the cache - how does it work?

[Florian Dittmer]

Hallo,

I find the following statement in the FAQ :

"If Borg detects that a repository has been modified since the local cache was 
updated it will need to rebuild the cache. This rebuild can be quite time 
consuming."

I have questions regarding this:

1. How exactly will Borg rebuild this cache?  Which data of the repository 
will Borg read and use to regenerate this cache? Only meta-data or the actual 
archived data? Will it re-use e.g. hash checksums from the archive's metadata 
or recalculate them on basis of actual data in the archive?

2. In a client/server setup, will the server or the client do the (if any) CPU 
intensive recalculation? Which (amount of) data will be transferred from 
server to client, in order to rebuild the cache? Only metadata , actual 
archive contents, or the server-side regenerated cache data?

3. In a client-only setup with archive storage on a mounted external network 
filesystem (e.g. sshfs), which (amount of) data will be transferred in order to 
rebuild the cache? Only metadata or actual archive contents?

Thank you very much

Florian