[SpamCop-List] Re: Why not allow bounces? They are required by RFC822!

[John E. Malmberg]

In article <dtktkd$dtn$1 at news.spamcop.net>,
  "Mike Easter" <MikeE at ster.invalid> writes:
> Frank Ellermann wrote:
>> Mike Easter wrote:
>>
>>> There shouldn't be any backscatter.
>
>> Your statement "there shouldn't be any backscatter" is _wrong_
>> without some qualifiers
>
> OK.

It is possible to create enterprise e-mail systems that generate no
backscatter even for users being over quota.

The vendors of such products should be strongly encouraged to update
their products for that to be the case.

Until their customers put pressure on them, though, they will continue
to design systems that produce backscatter.

The external mail server just needs a buffer for incoming e-mail for each
user which it is going to relay to an internal mail server.  That buffer
would have a small quota, and when it is full, e-mail gets rejected with
one of the SMTP reject codes as appropriate.

If an internal mail server refuses the relay, the incoming e-mail stays in
the external server until the condition with the internal mail server is
fixed, and if it takes too long, then the mail adminstrator for the domain
should be notified to take action.  No NDR is generated because the message
was received, and will be moved into the internal mail server once the
condition is cleared, and it is no different than an unread message in the
user's mailbox.

I would not be surprised if a lot of this could be done simply by using
special forwarding rules to send the current "backscatter" NDRs to a
system that uses a program to tell the external mail server to stop accepting
e-mail for the user over quota, and then use an over-quota privilege to
jam the last e-mail accepted into the user's mailbox on the internal server,
and send internal notifications to the applicable parties that can fix the
condition.

Backscatter from an internal system indicates an internal problem that
should not be foisted external systems.

Just because currently deployed systems can not eliminate backscatter
completely is not an excuse that systems should continue to be designed in such
a way that can cause backscatter.

-John
wb8tyw at qsl.network
Personal Opinion Only