[avahi] Is it time to drop .local SOA heuristic?

[Éric Bischoff]

Le vendredi 9 décembre 2022, 10:07:20 CET Esko Dijk a écrit :
> > Isn't that what the RFC 6762 recommends?
> 
> It is required (MUST). The optional thing the RFC states after this
> requirement:
 
>    Implementers MAY choose to look up such names concurrently via other
>    mechanisms (e.g., Unicast DNS) and coalesce the results in some
>    fashion.  

Sorry, I understood "MUST" too restrictively and missed this second paragraph. 
You are right.

> This is why I asked about whether a parallel lookup is possible; which would
> be according to the standard and would work for the particular use case of
> networks that have ".local" in use for unicast DNS.
> Unfortunately that
> doesn't work with nsswitch, if I understand Petr correctly. 

I always understood /etc/nsswitch.conf as trying one method of resolution 
after the other sequentially: for example first local files, then multicast 
DNS, then unicast DNS. No room for anything more complex. The only option is 
what to do in case of a negative answer, continue with next method or not.

As for your original question "whether (...) exist deployed .local zones in 
unicast DNS with a real data", I can confirm from personal experience that it 
is very common. Many people are seduced by the name "local" and use it for 
their local unicast DNS resolution.


Best,

-- 
Éric Bischoff