----- Original Message ----- From: "Fuzzy Fox" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Monday, July 14, 2003 7:43 AM Subject: [SAtalk] Re: Why wil "sa-learn" not learn?
> Simon Byrnand <[EMAIL PROTECTED]> wrote: > > > * Are you learning a proportionate amount of ham as well ? [...] > > * Bayes doesn't "learn" particular messages, it learns the statistics > > of the words used in the messages. [...] > > None of this addresses the original posted concern: When you feed one or > more messages to sa-learn, it should report: "Learned from X messages", > where X is some number greater than zero. Indeed. :) Bayes is definitely working here when invoked via spamc/spamd. To give you an idea what I have in my logs: ------------------------------------------------ debug: bayes corpus size: nspam = 4410, nham = 2746 ... debug: Ran run_rbl_eval_test rule RCVD_IN_NJABL ======> got hit debug: Ran run_rbl_eval_test rule RCVD_IN_UNCONFIRMED_DSBL ======> got hit debug: Ran run_rbl_eval_test rule RCVD_IN_BL_SPAMCOP_NET ======> got hit debug: running meta tests; score so far=7.6 debug: auto-learn? safety=4, ham=1, spam=11, body-hits=7.6, head-hits=7.3 debug: auto-learn: currently using scoreset 3. recomputing score based on scoreset 1. debug: Score set 1 chosen. debug: auto-learn: original score: 10.9, recomputed score: 13.734 debug: Score set 3 chosen. debug: auto-learn? yes, spam (13.734 > 11) debug: Learning Spam debug: bayes: 79401 untie-ing debug: lock: 79401 created /usr/home/spamd/.spamassassin/bayes.lock.asarian-host.net.79401 ... debug: is spam? score=13.9 required=7 tests=BAYES_90,HTML_70_80,MIME_HTML_ONLY,OBFUSCATING_COMMENT,PYZOR_CHECK,RCV D_IN_BL_SPAMCOP_NET,RCVD_IN_NJABL,RCVD_IN_UNCONFIRMED_DSBL logmsg: identified spam (13.9/7.0) for user:2226 in 30.5 seconds, 1910 bytes. identified spam (13.9/7.0) for user:2226 in 30.5 seconds, 1910 bytes. ------------------------------------------------ > If sa-learn reports "Learned from 0 messages," it means that it didn't > put anything into the Bayes database. The main reason that I have seen > for this, is that SA thinks that it has already seen the message before, > meaning that the Message-ID in the mail is already stored in the bayes > database. Yes. And when that occurs, it will give me log-entries like: ------------------------------------------------ debug: [EMAIL PROTECTED]: already learnt correctly, not learning twice ------------------------------------------------ But, clearly, that is not the case here. > However, the original poster already stated that he used "sa-learn > --forget" to try to remove any traces of the message, and yet sa-learn > STILL continues to say "Learned from 0 messages", meaning that it does not > want to learn from the particular message. Yes. And I cannot see why it would do that. ------------------------------------------------ Learned from 0 messages. debug: lock: 21499 created /root/.spamassassin/bayes.lock.asarian-host.net.21499 debug: lock: 21499 trying to get lock on /root/.spamassassin/bayes with 0 retries debug: lock: 21499 link to /root/.spamassassin/bayes.lock: link ok debug: bayes: 21499 tie-ing to DB file R/W /root/.spamassassin/bayes_toks debug: bayes: 21499 tie-ing to DB file R/W /root/.spamassassin/bayes_seen ------------------------------------------------ There is only one difference: spamd runs as user "spamd", so its bayes_path is at: /usr/home/spamd/.spamassassin/ Instead of: /root/.spamassassin/ Which is used for "sa-learn" (as I type the latter from the shell-prompt, of course). But that path difference does not really exist, as both /usr/home/spamd/.spamassassin/ and /root/.spamassassin/ are in actuality symbolic links to /var/db/spamassassin/, where the Bayes database resides (and the databases and dir are owned by spamd). Thanks for with me, anyway. :) - Mark ------------------------------------------------------- This SF.Net email sponsored by: Parasoft Error proof Web apps, automate testing & more. Download & eval WebKing and get a free book. www.parasoft.com/bulletproofapps1 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
