"Murray S. Kucherawy" <[email protected]> wrote:

>The IESG has the following issue they'd like us to resolve on the DKIM
>reporting draft, and the result might also apply to the SPF reporting
>draft, before it can be approved for publication.
>
>"- is rr=all a good default? If a bad-actor sends a supposedly signed
>mail with a DKIM-Signature with 1000  fields then what happens if the
>Signer's DNS has no rr tag? Maybe some special case for unknown
>DKIM-Signature tokens or a max on the number of reports for a single
>message? (Note: I'm assuming that a separate report can be, or is to
>be, generated for each DKIM-Signature token that fails, clarifying that
>only one report is ever sent for one inbound mail with a broken
>signature would also fix this.)"
>
>It's my understanding that we would generate one report per failed
>signature, because a single message might bear multiple signatures from
>different domains that have different reporting requirements, so the
>parenthetical solution won't work for us.  A maximum on the number of
>reports per message is somewhat arbitrary given that a Verifier can do
>them in any order.
>
>We have a few options here:
>
>- require an "rr" tag, rather than having a default
>- have some default other than "all", chosen from the set of available
>report requests
>
>And we might in addition:
>
>- add a report request for unknown signature tags
>
>What does the WG think?  The telechat is this coming Thursday, so
>having this resolved by then would be ideal.

Since a domain doesn't have to publish a record asking for these reports and 
(as an added bonus) ra= doesn't have a default, there's no worry about reports 
being generated for domains that don't opt-in.

Typically, I think if I needed to moderate volume, rp= would be the right knob 
to turn anyway.

The most I would do is say something like SHOULD NOT generate more than one 
report to a single domain per message.  To the extent there's an issue here, I 
think that addresses it.

Scott K
_______________________________________________
marf mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/marf

Reply via email to