Path: utzoo!utgpu!watserv1!watmath!att!tut.cis.ohio-state.edu!pacific.mps.ohio-state.edu!zaphod.mps.ohio-state.edu!sunybcs!boulder!stan!dancer!imp From: imp@dancer.Solbourne.COM (Warner Losh) Newsgroups: comp.protocols.tcp-ip.domains Subject: Re: Can non-Internet site handle others domains? Message-ID: <1990Apr13.031029.22002@Solbourne.COM> Date: 13 Apr 90 03:10:29 GMT References: <141@dynasys.UUCP> Sender: news@Solbourne.COM Organization: Solbourne Computers Inc. Lines: 40 In article <141@dynasys.UUCP> jessea@dynasys.UUCP (Jesse W. Asher) writes: >To be more specific, the mechanism that provide you with mail forwarding >is the MX record, which has to point to a machine which has >an A record, like so In article jeremy@cs.swarthmore.edu (Jeremy Brest) writes: >Maybe I'm entering this discussion a little late, but why is this so? >It seems that when a mailer looks up a host and finds an MX record, it >should then start at the top again and look up that host, again to see >if it has an MX record. This couldn't be difficult to implement, so >why doesn't the protocol allow for this? You can have multiple MX records for a given host. The defined algorithm for choosing the order to try these records is something like: 1) Get all MX records for host n 2) Sort all MX records 3) If your local host is in the list, remove it and all other hosts that have a higher MX value. 4) Try the MX hosts in order from lowest to highest until the mail succeeds. Now, if you have to go out and get MX record for MX records for MX records, this entire process falls apart. It would be very easy to get mail looping because you have A MXing for B MXing for A, etc. The purpose of MX records is to tell mailers a *HOST* they can send your mail to, not some domain name that will handle the mail for you. If you try to expand them beyond this limited functionality, you will be creating problems for old implementations, as well as making non-trivial to find mail loops easier to come into existence. *THAT* is why it is hard to implement. Just my two cents from a former VMS/SMTP/DECNET MAILER hacker..... -- M. Warner Losh Solbourne Computer imp@solbourne.com (formerly warner@twg.com and warner@hydrovax.nmt.edu)