On 09/06/2026 19:03, Todd Herr wrote:
Forgive me, as I've not been aware of RFC 9824 prior to this thread, and I'm having a bit of trouble wrapping my head around it.

Am I to understand that RFC 9824 allows for the NODATA response to "[claim] that the name exists but doesn't have any data for the queried record type" but to include in that response the setting of the NXNAME bit to state that the claim that the name exists is false, because the name really doesn't exist? Do I have that right?

That's how it works. For example:

- "dig developers.cloudflare.com MX +dnssec" shows that the NXNAME bit is not set because the domain exists. - "dig doesnotexist.cloudflare.com MX +dnssec" shows that the NXNAME bit is set because the domain does not exist.

I believe NXDOMAIN can't be used here otherwise it would conflict with standard non-compact NSEC/NSEC3.


As RFC 9989 is currently written, a domain owner wanting "np" to apply for non-existent subdomains would seemingly want to ensure that they're not using Compact Denial of Existence for their domains at this time, as RFC 9989 makes it clear that non-existent domain in the DMARC context is consistent with the term's meaning in RFC 8020.

Alternatively, the domain owner can use the Compact Denial of Existence but also ensure that the sp or p tags have the same value as the np tag, so that if a non-existent domain isn't seen as such due to a 9824 response, a DMARC failure would still be treated as desired.

Both are valid workarounds, although the first one means disabling DNSSEC or changing authoritative DNS provider, which clearly isn't always a viable option (or not worth it), and the second one practically means not using the "np" tag (if both "sp" and "np" have the same value, you can just omit "np").

Most of all, few people realistically know they're using Compact Denial of Existence, and even less are aware of this limitation (I understand that it's the first time it's raised here).

It appears that other members of this WG have participated in dnsop, where RFC 9824 was born, so maybe someone else has more informed thoughts than us on this.

Thank you!

--
Matteo

_______________________________________________
dmarc mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to