NLSR

Using /localhop cannot reduce Nack-Duplicates in the case of ChronoSync Interests.

ScopeControl restricts /localhop Interest forwarding as follows:

• If PIT entry has at least one InRecord from a local face, it can be forwarded to local faces and non-local faces.
• If PIT entry has all InRecords from non-local faces, it can only be forwarded to local faces.

As soon as the local NLSR has expressed a sync Interest, NFD can forward a sync Interest received from one remote router to another remote router, and cause the same amount of Nack-Duplicates.

What's the rationale for this rule "If PIT entry has at least one InRecord from a local face, it can be forwarded to local faces and non-local faces."?

Answer to note-2:

The alternate is: "If PIT entry does not have any InRecord from a non-local face, it can be forwarded to non-local faces".

This breaks the scenario when two nodes try to send the same Interest to each other.

In what case will two nodes try to send the same Interest to each other using /localhop and why would you like to forward the interest to other nodes in this case?

Rules have changed: https://redmine.named-data.net/issues/3836

I have tested this on a 34 node topology, no Duplicate NACKs are generated by sync interests (this reduced Duplicate NACKs from ~47000 to ~15000, those generated are by LSA interests).