On May 26, 2012, at 2:32 PM, Jared Johnson wrote: >> Been there, tried that. I've had mx2.freebsd.org soft 4xx blocked for 1 >> day at a time for over a week. Besides the increased number of connections >> to my server, using a 4xx error makes almost no difference. As soon as >> the penalty box expires, the queued ham and spam pour in and mx2 gets >> tossed back into the penalty box. For most messages, a 4xx error ends up >> being the same as 5xx error except with a delay. > > I don't know if I'm typical, but I would personally be very hesitant to > reject or even defer a server such as mx2.freebsd.org, which is known to > send legitimate mail,
Then we agree. I consider blocking mail from even "mostly good" servers to be a bug. Having to discover this and manually whitelist to prevent it is also a bug. Which is why I'm exploring a "karma" based solution instead. > controlled by spammers or botnets is not going to > retry messages indefinitely like this, so it should be much more > susceptible to a 'penalty box' regardless of whether you reject or defer. In my observations, there's been very little difference in appearance between a good server innocently retrying, and a naughty server trying a new spam. Matt