I've recently turned to rspamd 1.7.1 as my antispam solution. So far, it seems to be effective, but I've only been using it for a day or so. Interestingly, the latest version offers neural network-based analysis, which is probably the right direction for development to be taking (as one factor in making the classification decision). I also found that I could eliminate a lot of undesirable SMTP connections by setting reject_unknown_client_hostname in the Postfix configuration. My host is well connected with good DNS resolution, so I assume that any errors in this regard are probably due to the client host's not being a legitimate mail server. Comments and suggestions for further improvement are welcome.