Thanks Mike. Roughly what percentage of spam gets through?
I am a bit worried about blocking people with dynamic IP addresses say from their ISP, if they "inherit" an IP address recently used by an infected PC they will still be in the RBL and get blocked. Do you get many problems like that? Is it a good idea to block them so early, or should I wait and use the RBL's to score them in SA later in the procmail delivery. Obviously later will use up more CPU but will I get less false rejections? Greylisting seemed to be a better compromise, it does not reject anything it just adds a delay, this seems better. What do you think? Thanks Matthew -----Original Message----- From: Mike Jackson [mailto:[EMAIL PROTECTED] Sent: 29 January 2007 13:08 To: [EMAIL PROTECTED] Subject: Re: Should I use greylisting > So in your opinion, what is the best way to reject spam early in the mail > delivery, in order to reduce the load on spam assassin. Here's my anti-spam chain: 1. RBLs. These are the ones I use, in order: zen.spamhaus.org dynablock.njabl.org dsn.rfc-ignorant.org bogusmx.rfc-ignorant.org bl.spamcop.net 2. SPF milter. But, this blocks very little mail. In fact, so little that I'm wondering why I bother other than that all the cool kids are supporting SPF. 3. SpamAssassin invoked from procmail. I keep my Bayes database well-fed. I use a few rulesets from rulesemporium.com kept up to date with rules_du_jour, but they're not as effective as one would hope. I use razor, though only for checking (I don't report anything at this point). I have the ImageInfo plugin, but it doesn't seem to catch much either.
