Re: [DNSOP] On obsoleting DNSSEC RFCs; Example from RFC 4035

[Joe Abley <[e-mail]>]

Hi Дилян,

On 11 Nov 2019, at 14:07, Дилян Палаузов <[e-mail]> wrote:

> Why the example from RFC 4035 Appendix A generats RRSIG for DNSKEY signed by the ZSK, but for IANA.ORG there is no such
> RRSIG?

Some signing toolchains sign every RRSet in the a zone with the ZSK and additionally sign the DNSKEY RRSet with a KSK (in situations where both ZSK and KSK are used). Some toolchains specifically suppress the signature over the DNSKEY RRSet. Some toolchains give you one of those as a default with an option to switch to the other behaviour.

Using ZSKs and KSKs is conventional and often assumed, but perhaps it's reasonable for some implementers to worry about making assumptions and to err on the side of too many signatures. It's might not always be possible to infer intent from a single SEP bit on one RR. In a conventional scenario the signature by the ZSK over the DNSKEY RRSet is not needed, since for the ZSK to be useful the KSK must already be trusted, and the mechanism for gaining trust in a published ZSK is to verify the signature by a trusted KSK over the DNSKEY RRSet. The ZSK signature adds nothing useful.

More generally, I suppose, this is an example of the specification not demanding a particular behaviour, and implementers having the freedom to make their own decisions. In this case I can't think of a reason why the ambiguity is harmful, though.

> If this is not the right mailing list, which is the right one?

I think this is a perfectly good place to ask that question.


Joe