On Mon, Aug 17, 2009 at 5:17 AM, Matus UHLAR - fantomas<uh...@fantomas.sk> wrote:
> Did you properly configure trusted_networks and internal_networks? Yes, I double-checked that. I've read and re-read various docs on trusted_networks and internal_networks and I'm *pretty* sure I've got it right. :-) I also tried many different configurations of trusted/internal in case that made a difference, but none of them seemed to. > Did you edit all *.pre files to load plugins you need? Yes. I had a problem at first with the .pre files for 3.2.5 not being installed, but corrected that. Debug showed everything loading that I would expect to load. > Did you change scores for some rulesets in 3.1.9? Did you comment that in > 3.2.5? I did change scores for some rulesets in 3.1.9, but those changes did seem to hit just fine in 3.2.5. Charles Gregory wrote: > When you compare the hits on a message, is there any difference in BAYES? Unfortunately, no: in fact, Bayes was often the only thing that was hitting in 3.2.5. > What rules do you see triggering on 3.1.9 that are being missed on 3.2.5? Er. Good question. It *seemed* as though almost nothing was hitting on 3.2.5 but Bayes and a few custom rules, but I didn't run both in parallel to get a list of specific rules that one or the other hit or missed. (By that time, we were all "Oh, God, make the spam stop!" heh) Since no one else seems to have had the same experience, I guess I'll throw 3.2.5 on a non-critical server and see if I can't test the same corpus of spam against it and 3.1.9. Thanks! K.
