I'm using bayes (spamassassin 2.63) in a site wide mode on a fairly heavy volume server (~100,000 msgs/day) and I'm using auto learn and I do a manual rebuild at regular intervals to avoid excessive expiry times and it all seems to be working fine with my bayes_toks database staying between 10MB and 20MB.
My only problem is that the bayes_seen database just keeps getting bigger and bigger, and I don't know what to do about it. Its currently 160MB and growing. I'm a bit reluctant to delete it because I do use salearn to re-learn messages from time to time and I figure it must be needed for that, and I'm not sure what else its used for. I'm curious to know what other people do to keep the size of the bayes_seen database in check. Should I just delete it from time to time? - rick
