DNSWL Email Authentication Method Extension

One of the many checks that mail servers carry out is to query DNS whitelists (DNSWL, ). The semantics of DNSWL is similar to that of Vouch By Reference (VBR, ); that is, an external organization, trusted by the receiving mail transfer agent (MTA), vouches for the sender. Differently from VBR, the lookup is based on the IP address. That allows it to occur very early in an SMTP transaction, and thus it can be used to counterweight policies that typically occur at those early stages too, such as the Sender Policy Framework (SPF, ). Nevertheless, the result of a DNSWL lookup is likely used at later stages as well; for example, a delivery agent can use it to estimate the spamminess of an email message. Reusing the previously obtained result is more consistent than issuing multiple queries, and saves resources.

In this document, the acronym DNSWL is used both to refer to a generic organization that publishes a whitelist, and to indicate the use of such service, that is the method. The organization defines the DNS zone and the record type(s) to be queried, as well as the meaning of the listings and the procedures to maintain them. Unlike other methods, such as dkim or spf, there is little or no work to be carried out by the receiving MTA on the values retrieved from the DNS query. As explained in Section 2.3 of , DNSWLs encode taxonomical details about the mail sender as bit masks of type A records. The receiving MTA just needs to work out whether the data returned is at all applicable. In that case, it uses that data as an outsourced extension of its local policy. Thus, in a broad sense, the result of the method is the content returned by the query: The name of the DNSWL, which defines the meaning of the result. If the MTA uses a private mirror, this can be different from the zone actually queried. The bit mask value. This entry ("propspec" in the grammar of ) can be repeated if the DNSWL returns multiple A records. The TXT record, or the relevant part of it, if any. This can contain a domain name or an abuse reporting address, which can be used as described in Section 5.3 of . In a strict sense, the result of the method states how the query did: The query successfully returned applicable records. The query worked, but no applicable record was found. There was a transient problem with the query. There is a setup problem, the query cannot work.

There is a registry of Email Authentication Methods created by RFC5451. The method described in this document is referred by , it has three ptype.Property values detailed in .

Method Defined version dnswl [this rfc??] 1 ptype Property Value Status dns zone The origin of the results active policy ip type A result active policy txt type TXT result active In addition, this method reuses four of the values already defined in the Email Authentication Result Names associated registry. They are listed in .

Code Meaning Status pass active fail active temperror active permerror active

[Note to RFC Editor: please remove this entire section before publication.] This section records the status of a known implementation of the method described in this document at the time of writing, based on a proposal described in . See that document for further boilerplate. Courier-MTA is a full-featured, mature mail server, first publicly released in May 2000. A beta release in February 2013 introduced Authentication-Results in combination with DNS-based whitelists. This document is based on that implementation, which made it to production release 0.71, after one month testing. End-user documentation of that feature is available online at http://www.courier-mta.org/couriertcpd.html#idp5867072 In prior releases, only the -block option was present, and the Authentication-Results header field was handled by add-ons, not by the core implementation. The -allow option was added so that black and white lists can be configured using mostly symmetrical syntax. Finally, an option was added to inhibit SPF reject-on-fail for whitelisted senders. It is not possible to know how many installations of Courier-MTA have enabled these new features. However, no questions have been asked about them on the mailing list, yet. Despite the amount of spam, there seems to be little traction for this kind of development. The only DNSWL known to have been used for this purpose is dnswl.org. See http://www.dnswl.org/. With nearly 150K entries, it can make the email messages that get at least one authentication "pass" overreach a critical mass: It seems that subscribing to that list is easier, for some mail admins, than implementing other authentication methods.

All of the considerations described in Section 8 of apply. In addition, the usual caveats apply about importing text from external online sources. Although queried DNSWLs are well known, trusted entities, it is suggested that TXT records be reported only if, upon inspection, their content is deemed actually actionable. If they contain non-ASCII characters, they need to be encoded as appropriate.