> Any suggestions would be appreciated since we would really > like to start > using the SQL Bayes feature ( - we had this turned off.)
We switched to 3.0pre when we started to have a load issue with 2.63. However, it didn't initially appear any better, even after going back to the bare minimum tests. What cured it in the end for us was switching spamd to using UNIX domain sockets rather than tcp/ip. It's now running better than ever, even with more SARE rulesets than before. Chris Hine
