We could, for catchup, simply send over a manifest of a downstream librarian's stores to an upstream librarian, and have them entered as remote instances of those files (including checksum checking).
So a procedure would look like this:
Librarian A dumps a manifest_A
Files and manifest_A copied
Files ingested into Librarian B using manifest_A
Librarian B dumps manifest_B
Manifest_B is copied to Librarian A host
Librarian A 'ingests' manifest_B as a remote instance table for Librarian B
We could, for catchup, simply send over a manifest of a downstream librarian's stores to an upstream librarian, and have them entered as remote instances of those files (including checksum checking).
So a procedure would look like this: