http://issues.apache.org/SpamAssassin/show_bug.cgi?id=3109
------- Additional Comments From [EMAIL PROTECTED] 2006-04-20 21:51 ------- maybe it depends what you are short-circuiting. since all your sc tests rely on bayes at least, that may be something that is pushing out your scan times? as you can see here, my spam:ham ratio is pretty close yesterday, yet the time my system spent processing it was 2/5th of how long the ham took. Spam: 8140 Time Spent Processing Spam: 2.88 hours Ham: 9073 Time Spent Processing Ham: 7.30 hours i dont have any s/c ham rules expect for whitelists, so you see a pretty small sample size of sc=ham below. # cat /var/log/spamassassin/spamd.log | perl test.pl sc=spam: 4085 messages, total time 2346.59999999995, avg 0.574443084455313 sc=no: 6064 messages, total time 18832.1, avg 3.1055573878628 sc=ham: 134 messages, total time 68, avg 0.507462686567164 sc=all: 10283 messages, total time 21246.7000000002, avg 2.06619663522321 as you can see, my sc=spam and sc=ham avg scantimes are 6x faster than those message that were not s/c. how are your scantimes 4-8 seconds? old hardware? although i have yet to put your new implementation into production, i'm still using the hacked HARVEST_DNSBL_PRIORITY. as soon as i get some time to do testing on the new stuff i'll provide some more data. ------- You are receiving this mail because: ------- You are the assignee for the bug, or are watching the assignee.
