When enabled (default is disabled), domains with insecure MX selection but valid DANE TLSA records use a distinct site name. This prevents an insecure domain from denying DANE authentication to DNSSEC secure sibling domains on the same MX hosts.
2.2 KiB
set_dane_enabled
kumo.dns.set_dane_enabled(ENABLED)
{{since('dev')}}
Controls whether MX resolution probes
DANE TLSA records to
influence the site_name for a domain. ENABLED is a boolean. The default
value is false.
Enable it in deployments that deliver with DANE:
kumo.on('pre_init', function()
kumo.dns.set_dane_enabled(true)
end)
This is distinct from the per-egress-path enable_dane option, which governs whether DANE is enforced when connecting. In short:
kumo.dns.set_dane_enabled: whether resolution probesTLSArecords and lets them modify the site_name.enable_dane(egress path): whether DANE authenticates the peer when connecting.
Why this is a global option
Resolution assigns the site_name before the egress path is selected, so no
per-path enable_dane value exists yet for it to read.
How DANE influences the site_name
A domain is given a distinct site_name only when its own MX selection was not
DNSSEC-secure, yet one of its MX hosts publishes DANE TLSA records. Such a
domain resolves to a site_name prefixed with [dane-insecure]. Since the
site_name is the ready-queue key, the prefixed domain gets a ready queue of
its own.
Without the tag, this insecure domain could share a ready queue with an untagged, DNSSEC-secure sibling domain that resolves to the same MX hosts. A connection the insecure domain opened on that shared queue would not be DANE-authenticated, even though the secure sibling's policy requires it.
The remaining cases are left untagged and keep sharing the common ready queue:
- A domain whose MX selection is DNSSEC-secure is never tagged.
- An insecure domain whose MX hosts don't publish
TLSArecords at all (the common case, including domains hosted on providers that do not deploy DANE) is never tagged. - If the
TLSAlookup cannot be completed, the domain is left untagged rather than split off.
Like DANE enforcement itself, this requires a DNSSEC-validating resolver. See
enable_dane. Without one, TLSA
records never validate as secure and no domain is ever tagged.