en-US/about_MXRecords.help.txt

TOPIC
   about_MXRecords


Network Working Group Craig Partridge
Request for Comments: 974 CSNET CIC BBN Laboratories Inc
                                                            January 1986

                   MAIL ROUTING AND THE DOMAIN SYSTEM


Status of this Memo

   This RFC presents a description of how mail systems on the Internet
   are expected to route messages based on information from the domain
   system described in RFCs 882, 883 and 973. Distribution of this memo
   is unlimited.

Introduction

   The purpose of this memo is to explain how mailers are to decide how
   to route a message addressed to a given Internet domain name. This
   involves a discussion of how mailers interpret MX RRs, which are used
   for message routing. Note that this memo makes no statement about
   how mailers are to deal with MB and MG RRs, which are used for
   interpreting mailbox names.

   Under RFC-882 and RFC-883 certain assumptions about mail addresses
   have been changed. Up to now, one could usually assume that if a
   message was addressed to a mailbox, for example, at LOKI.BBN.COM,
   that one could just open an SMTP connection to LOKI.BBN.COM and pass
   the message along. This system broke down in certain situations,
   such as for certain UUCP and CSNET hosts which were not directly
   attached to the Internet, but these hosts could be handled as special
   cases in configuration files (for example, most mailers were set up
   to automatically forward mail addressed to a CSNET host to
   CSNET-RELAY.ARPA).

   Under domains, one cannot simply open a connection to LOKI.BBN.COM,
   but must instead ask the domain system where messages to LOKI.BBN.COM
   are to be delivered. And the domain system may direct a mailer to
   deliver messages to an entirely different host, such as SH.CS.NET.
   Or, in a more complicated case, the mailer may learn that it has a
   choice of routes to LOKI.BBN.COM. This memo is essentially a set of
   guidelines on how mailers should behave in this more complex world.

   Readers are expected to be familiar with RFCs 882, 883, and the
   updates to them (e.g., RFC-973).









Partridge [Page 1]
 


RFC 974 January 1986
Mail Routing and the Domain System


What the Domain Servers Know

   The domain servers store information as a series of resource records
   (RRs), each of which contains a particular piece of information about
   a given domain name (which is usually, but not always, a host). The
   simplest way to think of a RR is as a typed pair of datum, a domain
   name matched with relevant data, and stored with some additional type
   information to help systems determine when the RR is relevant. For
   the purposes of message routing, the system stores RRs known as MX
   RRs. Each MX matches a domain name with two pieces of data, a
   preference value (an unsigned 16-bit integer), and the name of a
   host. The preference number is used to indicate in what order the
   mailer should attempt deliver to the MX hosts, with the lowest
   numbered MX being the one to try first. Multiple MXs with the same
   preference are permitted and have the same priority.

   In addition to mail information, the servers store certain other
   types of RR's which mailers may encounter or choose to use. These
   are: the canonical name (CNAME) RR, which simply states that the
   domain name queried for is actually an alias for another domain name,
   which is the proper, or canonical, name; and the Well Known Service
   (WKS) RR, which stores information about network services (such as
   SMTP) a given domain name supports.

General Routing Guidelines

   Before delving into a detailed discussion of how mailers are expected
   to do mail routing, it would seem to make sense to give a brief
   overview of how this memo is approaching the problems that routing
   poses.

   The first major principle is derived from the definition of the
   preference field in MX records, and is intended to prevent mail
   looping. If the mailer is on a host which is listed as an MX for the
   destination host, the mailer may only deliver to an MX which has a
   lower preference count than its own host.

   It is also possible to cause mail looping because routing information
   is out of date or incomplete. Out of date information is only a
   problem when domain tables are changed. The changes will not be
   known to all affected hosts until their resolver caches time out.
   There is no way to ensure that this will not happen short of
   requiring mailers and their resolvers to always send their queries to
   an authoritative server, and never use data stored in a cache. This
   is an impractical solution, since eliminating resolver caching would
   make mailing inordinately expensive. What is more, the out-of-date
   RR problem should not happen if, when a domain table is changed,


Partridge [Page 2]
 


RFC 974 January 1986
Mail Routing and the Domain System


   affected hosts (those in the list of MXs) have their resolver caches
   flushed. In other words, given proper precautions, mail looping as a
   result of domain information should be avoidable, without requiring
   mailers to query authoritative servers. (The appropriate precaution
   is to check with a host's administrator before adding that host to a
   list of MXs).

   The incomplete data problem also requires some care when handling
   domain queries. If the answer section of a query is incomplete
   critical MX RRs may be left out. This may result in mail looping, or
   in a message being mistakenly labelled undeliverable. As a result,
   mailers may only accept responses from the domain system which have
   complete answer sections. Note that this entire problem can be
   avoided by only using virtual circuits for queries, but since this
   situation is likely to be very rare and datagrams are the preferred
   way to interact with the domain system, implementors should probably
   just ensure that their mailer will repeat a query with virtual
   circuits should the truncation bit ever be set.

Determining Where to Send a Message

   The explanation of how mailers should decide how to route a message
   is discussed in terms of the problem of a mailer on a host with
   domain name LOCAL trying to deliver a message addressed to the domain
   name REMOTE. Both LOCAL and REMOTE are assumed to be syntactically
   correct domain names. Furthermore, LOCAL is assumed to be the
   official name for the host on which the mailer resides (i.e., it is
   not a alias).

Issuing a Query

   The first step for the mailer at LOCAL is to issue a query for MX RRs
   for REMOTE. It is strongly urged that this step be taken every time
   a mailer attempts to send the message. The hope is that changes in
   the domain database will rapidly be used by mailers, and thus domain
   administrators will be able to re-route in-transit messages for
   defective hosts by simply changing their domain databases.

   Certain responses to the query are considered errors:

      Getting no response to the query. The domain server the mailer
      queried never sends anything back. (This is distinct from an
      answer which contains no answers to the query, which is not an
      error).

      Getting a response in which the truncation field of the header is



Partridge [Page 3]
 


RFC 974 January 1986
Mail Routing and the Domain System


      set. (Recall discussion of incomplete queries above). Mailers
      may not use responses of this type, and should repeat the query
      using virtual circuits instead of datagrams.

      Getting a response in which the response code is non-zero.

   Mailers are expected to do something reasonable in the face of an
   error. The behaviour for each type of error is not specified here,
   but implementors should note that different types of errors should
   probably be treated differently. For example, a response code of
   "non-existent domain" should probably cause the message to be
   returned to the sender as invalid, while a response code of "server
   failure" should probably cause the message to be retried later.

   There is one other special case. If the response contains an answer
   which is a CNAME RR, it indicates that REMOTE is actually an alias
   for some other domain name. The query should be repeated with the
   canonical domain name.

   If the response does not contain an error response, and does not
   contain aliases, its answer section should be a (possibly zero
   length) list of MX RRs for domain name REMOTE (or REMOTE's true
   domain name if REMOTE was a alias). The next section describes how
   this list is interpreted.

Interpreting the List of MX RRs

   NOTE: This section only discusses how mailers choose which names to
   try to deliver a message to, working from a list of RR's. It does
   not discuss how the mailers actually make delivery. Where ever
   delivering a message is mentioned, all that is meant is that the
   mailer should do whatever it needs to do to transfer a message to a
   remote site, given a domain name for that site. (For example, an
   SMTP mailer will try to get an address for the domain name, which
   involves another query to the domain system, and then, if it gets an
   address, connect to the SMTP TCP port). The mechanics of actually
   transferring the message over the network to the address associated
   with a given domain name is not within the scope of this memo.

   It is possible that the list of MXs in the response to the query will
   be empty. This is a special case. If the list is empty, mailers
   should treat it as if it contained one RR, an MX RR with a preference
   value of 0, and a host name of REMOTE. (I.e., REMOTE is its only
   MX). In addition, the mailer should do no further processing on the
   list, but should attempt to deliver the message to REMOTE. The idea




Partridge [Page 4]
 


RFC 974 January 1986
Mail Routing and the Domain System


   here is that if a domain fails to advertise any information about a
   particular name we will give it the benefit of the doubt and attempt
   delivery.

   If the list is not empty, the mailer should remove irrelevant RR's
   from the list according to the following steps. Note that the order
   is significant.

      For each MX, a WKS query should be issued to see if the domain
      name listed actually supports the mail service desired. MX RRs
      which list domain names which do not support the service should be
      discarded. This step is optional, but strongly encouraged.

      If the domain name LOCAL is listed as an MX RR, all MX RRs with a
      preference value greater than or equal to that of LOCAL's must be
      discarded.

   After removing irrelevant RRs, the list can again be empty. This is
   now an error condition and can occur in several ways. The simplest
   case is that the WKS queries have discovered that none of the hosts
   listed supports the mail service desired. The message is thus deemed
   undeliverable, though extremely persistent mail systems might want to
   try a delivery to REMOTE's address (if it exists) before returning
   the message. Another, more dangerous, possibility is that the domain
   system believes that LOCAL is handling message for REMOTE, but the
   mailer on LOCAL is not set up to handle mail for REMOTE. For
   example, if the domain system lists LOCAL as the only MX for REMOTE,
   LOCAL will delete all the entries in the list. But LOCAL is
   presumably querying the domain system because it didn't know what to
   do with a message addressed to REMOTE. Clearly something is wrong.
   How a mailer chooses to handle these situations is to some extent
   implementation dependent, and is thus left to the implementor's
   discretion.

   If the list of MX RRs is not empty, the mailer should try to deliver
   the message to the MXs in order (lowest preference value tried
   first). The mailer is required to attempt delivery to the lowest
   valued MX. Implementors are encouraged to write mailers so that they
   try the MXs in order until one of the MXs accepts the message, or all
   the MXs have been tried. A somewhat less demanding system, in which
   a fixed number of MXs is tried, is also reasonable. Note that
   multiple MXs may have the same preference value. In this case, all
   MXs at with a given value must be tried before any of a higher value
   are tried. In addition, in the special case in which there are
   several MXs with the lowest preference value, all of them should be
   tried before a message is deemed undeliverable.



Partridge [Page 5]
 


RFC 974 January 1986
Mail Routing and the Domain System


Minor Special Issues

   There are a couple of special issues left out of the preceding
   section because they complicated the discussion. They are treated
   here in no particular order.

   Wildcard names, those containing the character '*' in them, may be
   used for mail routing. There are likely to be servers on the network
   which simply state that any mail to a domain is to be routed through
   a relay. For example, at the time that this RFC is being written, all
   mail to hosts in the domain IL is routed through RELAY.CS.NET. This
   is done by creating a wildcard RR, which states that *.IL has an MX
   of RELAY.CS.NET. This should be transparent to the mailer since the
   domain servers will hide this wildcard match. (If it matches *.IL
   with HUJI.IL for example, a domain server will return an RR
   containing HUJI.IL, not *.IL). If by some accident a mailer receives
   an RR with a wildcard domain name in its name or data section it
   should discard the RR.

   Note that the algorithm to delete irrelevant RRs breaks if LOCAL has
   a alias and the alias is listed in the MX records for REMOTE. (E.g.
   REMOTE has an MX of ALIAS, where ALIAS has a CNAME of LOCAL). This
   can be avoided if aliases are never used in the data section of MX
   RRs.

   Implementors should understand that the query and interpretation of
   the query is only performed for REMOTE. It is not repeated for the
   MX RRs listed for REMOTE. You cannot try to support more extravagant
   mail routing by building a chain of MXs. (E.g. UNIX.BBN.COM is an MX
   for RELAY.CS.NET and RELAY.CS.NET is an MX for all the hosts in .IL,
   but this does not mean that UNIX.BBN.COM accepts any responsibility
   for mail for .IL).

   Finally, it should be noted that this is a standard for routing on
   the Internet. Mailers serving hosts which lie on multiple networks
   will presumably have to make some decisions about which network to
   route through. This decision making is outside the scope of this
   memo, although mailers may well use the domain system to help them
   decide. However, once a mailer decides to deliver a message via the
   Internet it must apply these rules to route the message.









Partridge [Page 6]
 


RFC 974 January 1986
Mail Routing and the Domain System


Examples

   To illustrate the discussion above, here are three examples of how
   mailers should route messages. All examples work with the following
   database:

      A.EXAMPLE.ORG IN MX 10 A.EXAMPLE.ORG
      A.EXAMPLE.ORG IN MX 15 B.EXAMPLE.ORG
      A.EXAMPLE.ORG IN MX 20 C.EXAMPLE.ORG
      A.EXAMPLE.ORG IN WKS 10.0.0.1 TCP SMTP

      B.EXAMPLE.ORG IN MX 0 B.EXAMPLE.ORG
      B.EXAMPLE.ORG IN MX 10 C.EXAMPLE.ORG
      B.EXAMPLE.ORG IN WKS 10.0.0.2 TCP SMTP

      C.EXAMPLE.ORG IN MX 0 C.EXAMPLE.ORG
      C.EXAMPLE.ORG IN WKS 10.0.0.3 TCP SMTP

      D.EXAMPLE.ORG IN MX 0 D.EXAMPLE.ORG
      D.EXAMPLE.ORG IN MX 0 C.EXAMPLE.ORG
      D.EXAMPLE.ORG IN WKS 10.0.0.4 TCP SMTP

   In the first example, an SMTP mailer on D.EXAMPLE.ORG is trying to
   deliver a message addressed to A.EXAMPLE.ORG. From the answer to its
   query, it learns that A.EXAMPLE.ORG has three MX RRs. D.EXAMPLE.ORG
   is not one of the MX RRs and all three MXs support SMTP mail
   (determined from the WKS entries), so none of the MXs are eliminated.
   The mailer is obliged to try to deliver to A.EXAMPLE.ORG as the
   lowest valued MX. If it cannot reach A.EXAMPLE.ORG it can (but is
   not required to) try B.EXAMPLE.ORG. and if B.EXAMPLE.ORG is not
   responding, it can try C.EXAMPLE.ORG.

   In the second example, the mailer is on B.EXAMPLE.ORG, and is again
   trying to deliver a message addressed to A.EXAMPLE.ORG. There are
   once again three MX RRs for A.EXAMPLE.ORG, but in this case the
   mailer must discard the RRs for itself and C.EXAMPLE.ORG (because the
   MX RR for C.EXAMPLE.ORG has a higher preference value than the RR for
   B.EXAMPLE.ORG). It is left only with the RR for A.EXAMPLE.ORG, and
   can only try delivery to A.EXAMPLE.ORG.

   In the third example, consider a mailer on A.EXAMPLE.ORG trying to
   deliver a message to D.EXAMPLE.ORG. In this case there are only two
   MX RRs, both with the same preference value. Either MX will accept
   messages for D.EXAMPLE.ORG. The mailer should try one MX first (which
   one is up to the mailer, though D.EXAMPLE.ORG seems most reasonable),
   and if that delivery fails should try the other MX (e.g.
   C.EXAMPLE.ORG).


Partridge [Page 7]