I think this is a duplicate of #6644 and #7244 right?

I think this is a duplicate of #6644 and #7244 right?

It looks similar, but I think my case is distinct.

I have BIND zone file for my domain (example.com) under version control.

I add a separate zone file containing a TXT record for ACME challenge (_acme-challenge.example.com) and set up keys and permissions in BIND to be able to modify it via rfc2136 protocol.

Since domain name in SOA record in BIND can't have underscores, I can't put _acme-challenge.example.com in SOA, and therefore certbot can't correctly deduce the domain from SOA.

Thanks for the additional info.

@adferrand or @joohoi, I think you both know more about this than me. When you get a few minutes, can you comment on this use case and whether or not it would be supported by the two PRs I reference above?

I think this situation is one of the pathological cases described by @hpax in #7244.

Here _acme-challenge.example.com is a zone on its own described out of the scope of the DNS server declared as SOA for the example.com zone. Typically a situation where recursive discovery will not work.

So I think having a flag to explicitly set the zone name is a good complement to override any auto discovery result if needed, or circumvent the inability to use the auto discovery.

Finally, if #7244 lands as a more generic implementation for any dns provider, this one should be generalized as well. If not, we can keep it for the sole purpose of rfc2136 plugin.