I received a good off-list reply that I'm responding to, here. I've anonymized it to protect the innocent.
> Ryan Thompson wrote: > > > With RBL checks enabled, spamc takes several seconds (7.8, on > > average) to process each message. I expect it to. However, the > > problem is that when run as a milter, the sending MTA is forced to > > wait, and, in some cases (i.e., when spamc needs somewhat longer > > than 7.8 seconds), this is causing timeouts with delivery, and > > consuming needless resources (sockets and RAM) to filter spam in > > "real time", when I'll never have any real-time status to report to > > the sender. [...] > > Ryan, > > I've offloaded the RBL into sendmail using it's built-in blacklist > features using the following in the sendmail.mc file. > > [...] Hello, Thanks for the tip. I'd like to benefit from SA's genetic algorithm, though, with respect to the RBL tests. Having sendmail reject anything from blacklisted hosts could lead to more false positives than I'd like to see (i.e., > 0 :-) > Takes the load off spamd significantly. I bet it would. Hmm. What'd be really useful is to have a switch to only use the RBL checks when necessary to classify a message. If the local tests already come up with a really high (or low) score (definable threshold in config, like the auto whitelist), don't bother with the RBL tests. The only time the RBL tests would be used is if the message score is already within a certain score interval. Something along those lines would certainly cut down on the amount of RBL queries generated, at what seems like a very small risk of incorrectly classifying a message. That still doesn't solve the slow MTA response problem I described in my original message, but it *would* improve overall performance. Then again, I'm not an SA developer, so I don't know exactly how feasable this would be. Thoughts? Thanks, - Ryan
