In article <407a44ce$0$430$8eec23a@newsreader.tycho.net>,
John DuBois <spcecdt@deeptht.armory.com> wrote:
>In article <Hw0Cyy.tvJ@wjv.com>, Bill Vermillion <bv@wjv.com> wrote:
>>In article <407907a7$0$435$8eec23a@newsreader.tycho.net>,
>>John DuBois <spcecdt@deeptht.armory.com> wrote:
>>>In article <HvyDBB.1rvv@wjv.com>, Bill Vermillion <bv@wjv.com> wrote:
>>>>nouser: /dev/null.
>>
>>>>Then you can alias the user to 'nouser'
>>
>>>The problem with this is that all such users will continue to be
>>>valid addresses forevermore. Anyone who sends legitimate mail to
>>>them will have their mail disappear into a black hole, spammers
>>>will send ever more traffic to them, mail generated automatically
>>>for those users will go unnoticed, etc. You really want to be
>>>able to mark all such addresses as no longer valid.

>>Spammers don't seem to pay any attention to anything
>>coming back, IF they even see it. So much I see comes from bad
>>addresses or forged addresses, that I've taken to throwing away
>>anything that's not right. It's not what I'm supposed to do but
>>the resources would be strained to the limits if I tried to bounce
>>all bad addresses back.

>You certainly shouldn't bounce mail except when absolutely unavoidable.

Absolutely.

>If your only choices are to bounce or drop in a black hole,
>dropping becomes a reasonable option. But that should only be the
>case if you use an MTA that does not verify addresses at SMTP
>time (an MTA that is in need of improvement!), or you have mail
>for your domain blindly forwarded to you by another host for some
>reason - for example, if you're at the end of a UUCP or other
>non-SMTP link.

Most of my mail work in the past 6 years has been handling mail
for 2 small ISPs. My problems come from the fact that these
machines are on 100Mbit ethernet links for less than 1000 feet
before they are connected to a global 40Gbs backbone. Being easy
to get to and with nothing between the spammer and my servers
makes them prime targets for spammers.

> It is much preferable for the SMTP servers for a domain to have
>knowledge of all addresses at that domain, so that attempts to
>send mail to nonexistant addresses can be rejected as soon as
>RCPT TO: is given. Unlike bouncing, this has some chance of
>actually affecting spammer address databases, and it minimizes
>resource consumption by preventing messages from being handed off
>in the first place. This is the model that all mail transport
>must move toward in this spammed age.

I do reject many up front. And thankfully all accounts are business
acounts with multiple users at the addresses with nothing being
passed off to other servers. I think I go far crazier than I alread
am working with typical end-users :-)

Bill


--
Bill Vermillion - bv @ wjv . com