> -----Original Message----- > From: John P Verel [mailto:jverel@xxxxxxxxxxxxx] > Sent: Monday, December 15, 2003 9:33 PM > To: Jake McHenry > Cc: shrike-list@xxxxxxxxxx > Subject: Re: Spam > > > > On 12/15/03 20:39 -0500, Jake McHenry wrote: > > I'm currently using spamassassin and mailscanner, working > well, but my > > boss doesn't want any spam to get through. I tried telling him that > > it's not going to catch everything, but he doesn't want to > listen. He > > told me to look into mimedefang, which his friend told him about. > > > > Does anyone know if mimedefang is any better or worse than > > spamassassin? > > Well, spamassassin does in fact 'de-fang' mime, so you've got > that already. > > My experience with spamassassin, which is EXCELLENT, is that > the key is to move all the false negatives to your spam > folder and to run the sa-learn program regularly. I've also > changed my rules set so that a Bayes90 hit gets an automatic > 5 points. I've had 1 false positive in the last month. > Spamassassin has caught between 650 and 700 in each of the > last three weeks, with I'd guess 10-15 false negatives a week > (e.g. 1 or 2 a day.) The number of false negatives keeps > dropping the more I use sa-learn. > > Spamassasin is the greatest, IMHO. > > John > Here is my output of sa-learn --dump [admin@ntlh admin]$ sa-learn --dump Use of uninitialized value in numeric lt (<) at /usr/lib/perl5/site_perl/5.8.0/Mail/SpamAssassin/BayesStore.pm line 1281. 0.000 0 0 0 non-token data: bayes db version 0.000 0 0 0 non-token data: nspam 0.000 0 0 0 non-token data: nham 0.000 0 0 0 non-token data: ntokens 0.000 0 0 0 non-token data: oldest atime 0.000 0 0 0 non-token data: current scan-count 0.000 0 0 0 non-token data: last expiry atime I'm running spamassassin 2.60 with mailscanner 4.23 I have bayes enabled in spam.assassin.prefs.conf in the mailscanner directory. It seems to be working for the most part, I'm getting probably around 80%. None is really being mis-marked, just missed. I also have bayes auto learn enabled in the config file, and in the maillog file, I can see the bayes engine is being used: BAYES_99 19,650 times in the log. Also using DCC and razor2. What can I do to get my percentage higher? Why isn't the spamassassin dump from above showing anything? I havn't used sa-learn before, I've just been submitting all missed mail to the razor2 servers. Thanks, Jake McHenry Nittany Travel MIS Coordinator http://www.nittanytravel.com -- Shrike-list mailing list Shrike-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/shrike-list
