On 08/12/2009 16:35, Charles Gregory wrote: > Sadly, with such a diverse user base, I cannot use a single Bayes DB > that would work well for all our users. My SMTP gateway (Mail Avenger) > works best if mail is scanned for *all* recipients, and so it is not > possible to use individual per-user Bayes. This is not an SA problem, > but just the nature of the SMTP gateway. It has to decide to accept or > reject the DATA transaction for ALL recipients. Once mail proves to be > lower scoring than the 10 threshold, individual user whitelists and > blacklists come into play, and other special per-user tests, but that > merely results in mail being diverted to their 'spamtrap' folder. I do > not 'bounce' mail once the SMTP gate is closed. :)
In cases were there is only a single recipient, I run SpamAssassin at SMTP time as the destination user. In cases where there are multiple recipients, it runs as the "nobody" user. This allows me to have per user preferences and bayes applied to the vast majority of incoming mail, during SMTP; only a tiny proportion of incoming mail here is multi-recipient... YMMV -- Mike Cardwell - IT Consultant and LAMP developer Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/ Technical Blog: https://secure.grepular.com/blog/
