Re: I-D Action:draft-klensin-rfc2821bis-10.txt

Paul Smith wrote:

> True. But if you could mandate that NOT having an MX record makes the 
> domain NOT valid for mail, then it might be easier to filter out some 
> junk. Also, it stops mail senders (sending bounce messages normally) 
> *arbitrarily* choosing to try to send the mail to a host.
> It's "arbitrary" because there's no way to set up a host name which 
> WON'T get mail (other than to do something extra such as set up a dummy 
> MX record, which would make anyone think that the name SHOULD get mail 
> contrary to the actual reason for doing it), so simply setting up a host 
> name causes the incorrect implication that you want to get mail on that 
> host.
I agree, but in my view, I don't think it is my place as a SMTP software 
provider to mandate it across the board.   It would have to be 
implementation option.
There is no question in my mind, if every (wide adoption) SMTP software 
began to do this, we would have an impact on both ends - good and bad.
But how much of an negative impact will it have and is it safe to assume 
those people are going to JUMP to fix their MX records?
Maybe they don't won't have a choice.

But in my view, again as a Software guy, I would leaned toward to 
maintaining compatibility and if we are asking for code change, then I 
prefer to look for something we can use to something extra that will 
justify a new change in "logic".   From this standpoint, I prefer to 
play the game safe technically and also legally.
For example, if the email had a DKIM signature, that is would be, IMV, a 
legal trigger to do new things that is above the normal legacy 
expectations.
> Yes, you can say that the lack of a waiting service port means it isn't 
> a mail host - but it DOESN'T. It means that it isn't a mail host *now*. 
> It might be in 2 minutes, or it might never be - there's no way of knowing.
Very true.  This is all part of an system retry logic and setup. For our 
setup/operation, a strong requirement is enabled for system availability.
The theory behind our CBV implementation is that if you're good and 
ready to send me mail, then you better be good and ready at the same 
precise moment of time to accept mail (verified by CBV).
> If there was some mandated, standard way of advertising that there 
> *should* be a service on a certain port, then that would work, but IMHO, 
> an MX record could do just that for port 25.
But MX is just a "menu" of A* records.  It is the A* (host) records that 
need to be resolved. Whether is a group or just 1 (implicit) host set, 
the real technical requirement can be stated as:
       A valid SMTP host is one with a A/AAAA record
       with a SMTP service port.

and the idea of system availability can be argued to be an out of scope 
policy question, not technical.
I think 2821bis covers this pretty well:

  4.5.4.2.  Receiving Strategy

   The SMTP server SHOULD attempt to keep a pending listen on the SMTP
   port (specified by IANA as port 25) at all times.  This requires the
   support of multiple incoming TCP connections for SMTP.  Some limit
   MAY be imposed but servers that cannot handle more than one SMTP
   transaction at a time are not in conformance with the intent of this
   specification.

   As discussed above, when the SMTP server receives mail from a
   particular host address, it could activate its own SMTP queuing
   mechanisms to retry any mail pending for that host address.

--
Sincerely

Hector Santos, CTO
http://www.santronics.com
http://santronics.blogspot.com