On Tue, 04 Feb 2025 at 11:27:59 +0100, Helmut Grohne wrote: > * While it would be possible to treat mDNS resolution independently > from mDNS publishing in the ballot, outcomes where they are handled > differently tend to cause more confusion and we rather want to have > them configured uniformly.
As evidence to confirm this as a good principle to follow, since I don't remember seeing anyone referencing the RFC in this discussion: mDNS doesn't really have resolvers and publishers as separate concepts, it just has *responders*, which do either or both of those things as required. The RFC explicitly recommends having at most one responder per machine: https://datatracker.ietf.org/doc/html/rfc6762#section-15 Avahi is a mDNS responder (that is its entire purpose). sd-resolved is sometimes a mDNS responder, among its other functions, and I think this decision is exactly about the circumstances under which sd-resolved's mDNS responder functionality should or should not be enabled. smcv (relevant hats: not a TC member; infrequent team-uploader of Avahi, but not actually in its Uploaders; co-maintainer of nss-mdns, which is an Avahi client; not a systemd maintainer.)
