En réponse à Theo Van Dinter <[EMAIL PROTECTED]>: > On Tue, Feb 25, 2003 at 01:46:19PM +0100, [EMAIL PROTECTED] > wrote: > > Looking at the bayes db it seems I have more than the 200 messages in > spam > > and ham categories required for bayes to kick in, however in the SA > reports > > I never see a BAYES_xx score... And this is based on more than 10000 > emails > > going through SA a day... > > If you pipe a message through "spamassassin -D", what is the debug > output? >
Given that my spamd script runs as user "filter" I had first to su to the user "filter" (otherwise spamassassin -D gives a " debug: bayes: no dbs present, cannot scan: /root/.spamassassin/bayes_toks "). The debug output of spamassassin proves that bayes kicks in for that sample run (and I see the appropriate BAYES_xx in the header report). Now the difference is that I do not run spamassassin, but spamc in my procmail recipe... And the options to spamd are "-d -L -x -u filter". Do you think the "-x" could cause this behaviour ? Also, running spamassassin -D did say something about expiring tokens, and it did expire like 20000 tokens !! The next test I did with spamassassin -D it did not expire any additional tokens (which is fine because it expired many of them at the previous run) -- this leads me to think that spamd does NOT expire tokens at all (otherwise I would not have 20000 to be expired), or is broken in some way ? Stephane ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
