[Pkg-utopia-maintainers] Bug#1077937: avahi-daemon: WARNING: Detected another IPv4 mDNS stack running on this host. This makes mDNS unreliable and is thus not recommended.

[Simon McVittie]

On Wed, 04 Sep 2024 at 18:09:55 +0200, Michael Biebl wrote:
> As I said, I prefer if systemd is built with -Ddefault-mdns=no as we have
> avahi which provides this kind of functionaliy and (experienced) users that
> want that kind of functionality from resolved can opt-in and enable it
> easily.
> One package (silently) disabling functionality of another package feels
> wrong to me.

I agree, it would seem strange to me to make avahi-daemon install a
systemd-resolved configuration snippet that disables sd-resolved's mDNS
support: that seems like a strange "action at a distance".

> Should one day, systemd-resolved supplant/replace avahi, then this a
> different story. But we are not there yet.

All of Michael's reasoning makes sense to me. I assume Fedora had similar
reasoning for why they're already doing what Michael is suggesting
we do in Debian. We have lots of packages that make use of Avahi APIs
specifically, and would not get equivalent functionality from sd-resolved
(e.g. cups, gvfs-backends, samba).

If at some future date we are in a position to remove Avahi from ordinary
desktop installations, have sd-resolved installed by default, and somehow
make sd-resolved responsible for the whole mDNS feature area, then that
would be fine; but my impression is that this is not (yet?) the case.

mDNS has two sides, resolving (getent hosts foo.local) and publishing
(advertising myhost.local and "My Service"._something._tcp.local to the
local network). Is there anything in systemd that intends to take over
the publishing side from Avahi, or is it only the resolving side?

    smcv