Re: [exim] hostname and HELO/EHLO response

Páxina inicial
Borrar esta mensaxe
Responder a esta mensaxe
Autor: Exim User's Mailing List
Data:  
Para: Exim User's Mailing List
Asunto: Re: [exim] hostname and HELO/EHLO response
[ On Friday, October 22, 2004 at 22:11:45 (+0100), Alan J. Flavell wrote: ]
> Subject: Re: [exim] hostname and HELO/EHLO response
>
> On Fri, 22 Oct 2004, Greg A. Woods wrote:
>
> > There's nothing like suddenly bouncing all mail from a broken site to
> > wake them up and get them to take action,
>
> I'm sure that would be true, *if* it could be achieved in a concerted
> fashion. But in our case we'll be told that we're the "only" site who
> has any problem with their mail, and one of our Professors will turn
> up to insist that his next 3 years funding, or latest and greatest
> research paper, is terminally at risk if we don't get our act together
> and ignore whatever offence the other site may be committing.
>
> It's a difficult tightrope to walk.


Indeed. :-)

I usually turn it around though and suggest to the local user that the
site he or she is having trouble with is the "only" such site we have
trouble with and that if they're the odd one out then it _must_ be
something to do with them, not us. (with a knowing wink of course)

Then I remind them that in the Public Internet today the responsibility
to thread mail through to the recipient is almost entirely on the
sender's side, especially given that the receiving side is already
successfully accepting tens, or hundreds, of thousands of other messages
from other sites every day, day in and day out.

Finally I offer to help to directly contact, and work with, the sender's
IT folks to resolve the problem if the user would be so kind as to have
his or her correspondent directly FAX us a printed copy of the bounce
message, complete with full headers.

Often this last request embarrasses the sender enough to cause them to
contact their own IT folks and the problem magically disappears a day or
so later. Sometimes though the underlying problem is that the sender is
too afraid (or too ignorant) to try to get support from his or her own
IT folks and is very glad to have a helping proxy work through things on
their behalf.

Through tactics such as this it seems I, or rather my front-line support
crew colleagues, have directly helped to spread a remarkable amount clue
around to a remarkably ignorant group of e-mail administrators around
the world. For example the support crew at a small regional ISP I help
out recently taught at least someone responsible for e-mail at Canada
Post, our equivalent of The Royal Mail, the importance of having and
using a proper postmaster alias (something I would have thought the post
office would have understood implicitly!). Unfortunately not all such
folks are so willing and able to learn the basics of Internet e-mail
despite the fact that they have responsibility for making it work for
their employers.

The reason these tactics are, as you say, like walking a tightrope, is
that the basic underlying premise of all SMTP reliability doesn't really
work very well at the moment. Ordinary people don't seem to make what
should be a very obvious real-world connection between a bounce message
and a returned piece of regular mail and thus they don't seem to
understand that bounce messages are very critically important signals to
the _sender_ that they have made a mistake somehow.

To come full circle though the proper response to the complaining user
is to remind them that just because someone can successfully send e-mail
to many other addresses doesn't mean they're going to be able to
successfully send mail to _all_ addresses _all_ of the time. Maybe they
(or their e-mail service provider) have not provided "sufficient
postage" (e.g. a working postmaster alias), or maybe they've screwed up
their addressing (e.g. they've used a bogus sender address or bogus
client hostname), or whatever. The point is that just because you drop
an envelope in a letter box doesn't mean it's going to get to who you
_think_ it should get to, and likewise in the e-world messages don't
magically get delivered 100% of the time unless _all_ the right bits and
pieces are _always_ in place and all the i's are dotted and t's are
crossed, etc.. Not every recipient has the same requirements -- in the
real world it takes more postage to send mail internationally with the
specific amount often depending on the exact destination, and in the
e-world some receiving sites are more strict about which other services
they accept messages from and under which conditions. Many more good
analogies can be made of course.

-- 
                        Greg A. Woods


+1 416 218-0098                  VE3TCP            RoboHack <woods@???>
Planix, Inc. <woods@???>          Secrets of the Weird <woods@???>