We did this at powweb.com fyi using postfix. i will try to dig up how we
implemented it. At the time it seemed pretty weak and that never really
changed imho. It seems to me it would be rather trival to make a spam
client listen on a few ports on their workstation rendering this whole
line of thinking useless pretty quickly.
Jesse
Matt Fretwell said:
> Marc Perkel wrote:
>
>> So - the more general question is - are there tests that we can do on
>> the connecting host that would provide useful information to determin if
>> they are a spammer or not a spammer. In this case I don't so much see an
>> opportunity to identify spam as I do to identify ham. So - responses
>> like "you'll be blocking me if you do that" are not relevant to this
>> idea.
>
>
> Firstly, those responses are perfectly relevant. However, (trying to
> subdue my usual helpful streak :), have you actually thought of putting an
> idea in to a limited production environment and actually checking the
> statistics? As twonky an idea as I still believe this is, no one can give
> you a firm idea on this due to the fact that it is an unknown. Systems
> vary. Setups vary. Workarounds vary.
>
> If you wish to find the probability of an unknown occuring within a given
> situation, you have two options. Speculation or statistics. One is
> useful. The other is not. Implement and measure. Obviously, with a
> situation like this, the higher the count with regards to mail passed,
> will give you an indication of the plausability of the notion. Conversely,
> a proactive blocking approach requires a higher figure with regards to
> mail rejected.
>
> You will simply have to test this out and figure out whether it is worth
> it.
>
>
> Matt
>
> --
> ## List details at http://www.exim.org/mailman/listinfo/exim-users
> ## Exim details at http://www.exim.org/
> ## Please use the Wiki with this list - http://www.exim.org/eximwiki/
>
-----------------
Jesse Geddis
LA Metro Wireless
-----------------
