> -----Original Message----- > From: Justin Mason [mailto:[EMAIL PROTECTED] > Sent: Thursday, October 09, 2003 12:20 PM > To: Mike Van Pelt > Cc: [EMAIL PROTECTED] > Subject: Re: [SAtalk] SA 2.60 *VERY* slow. > > > > Mike Van Pelt writes: > >This is nuts. I'm going to have to back out 2.60 tonight > >if I can't get this resolved today. (Which I really hate > >to do; 2.60 does seem to catch more spam.) > > It could be doing bayes rebuilds for some reason -- try > running "sa-learn --rebuild" before starting SpamAssassin. > > Also, getting a debug log from SpamAssassin would be > very helpful. >
I've noticed that the way bayes now expires has caused considerable load on my boxes now. I had to go in an hard code the 43200 numbers in BayesStore to 3600... otherwise my bayes database grows to 40 MB in 12 hours, and the continuous tie-ing looks to be the cause of the loads. running at 3600 seconds, my bayes db stays around 10-12mb, and everything is peachy. with 2.55, SA had a value that you could specify expire to run every 5000 messages, this option is no longer. as i get approx 20k messages an hour on a global bayes db, in 12 hours, i've learned way too many toks. making that 43200 number a configuration option might not be a bad idea? bayesstore with 43200 sec expiry requirement (22-40MB toks) causes 2.5-4.0 avg loads bayesstore with 3600 sec expiry requirement (10-12MB toks) causes 0.75-1.5 avg loads running without bayes gives me 0.4-0.8 avg loads dallas ------------------------------------------------------- This SF.net email is sponsored by: SF.net Giveback Program. SourceForge.net hosts over 70,000 Open Source Projects. See the people who have HELPED US provide better services: Click here: http://sourceforge.net/supporters.php _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
