Hi,
Having upgraded to 2.60 I've noticed that every mail that passes the SA
threshold (5.0 on my setup) always has BAYES_99...
5.4 BAYES_99 BODY: Bayesian spam probability is 99 to
100%
[score: 1.0000]
..in the report. Prior to the upgrade I had a variance in the BAYES_*
scores - as I would've expected. Now, it seems, with any SPAM-tagged
email its always 99-100% (which, ironically, is correct given its very
definitely spam)
I followed the normal procedure of "sa-learn --import" when upgrading to
2.60 (from 2.54), and my "sa-learn --dump magic" reads as follows:
bash-2.05b$ sa-learn --dump magic
0.000 0 2 0 non-token data: bayes db version
0.000 0 194809 0 non-token data: nspam
0.000 0 11332 0 non-token data: nham
0.000 0 136926 0 non-token data: ntokens
0.000 0 1064309780 0 non-token data: oldest atime
0.000 0 1065083380 0 non-token data: newest atime
0.000 0 1065078866 0 non-token data: last journal
sync atime
0.000 0 1065000935 0 non-token data: last expiry
atime
0.000 0 691200 0 non-token data: last expire
atime delta
0.000 0 110082 0 non-token data: last expire
reduction count
My .spamassassin directory structure now looks like this (post-upgrade):
total 52596
drwx------ 2 qmailq qmail 4096 Oct 2 09:29 .
drwxr-xr-x 18 qmailq qmail 4096 Oct 2 09:24 ..
-rw------- 1 qmailq qmail 41971712 Sep 23 09:51 auto-whitelist
-rw------- 2 qmailq qmail 1011832 Oct 2 09:29
auto-whitelist.dir
-rw------- 2 qmailq qmail 1011832 Oct 2 09:29
auto-whitelist.pag
-rw------- 1 qmailq qmail 82304 Oct 2 09:29 bayes_journal
-rw------- 1 qmailq qmail 21401600 Oct 2 09:29 bayes_seen
-rw------- 1 qmailq qmail 5111808 Oct 2 09:29 bayes_toks
-rw-r--r-- 1 qmailq qmail 1600 Apr 8 22:34 user_prefs
..prior to upgrade it looked like this (I have a backup of the
directory):
total 52496
drwx------ 2 qmailq qmail 4096 Sep 23 09:49 .
drwxr-xr-x 18 qmailq qmail 4096 Oct 2 09:24 ..
-rw------- 1 qmailq qmail 41971712 Sep 23 09:49 auto-whitelist
-rw------- 1 qmailq qmail 2343183 Sep 23 09:49 bayes_journal
-rw------- 1 qmailq qmail 4972 Sep 23 09:49 bayes_msgcount
-rw------- 1 qmailq qmail 20336640 Sep 23 09:48 bayes_seen
-rw------- 1 qmailq qmail 5423104 Sep 23 09:48 bayes_toks
-rw-r--r-- 1 qmailq qmail 1600 Apr 8 22:34 user_prefs
Does this appear correct? Why would my "bayes_journal" have been zero'd
- is this correct behaviour?
I've also noticed that my auto-whitelist isn't growing in size, and
"autolearn=no" is appearing on emails, despite having:
auto_learn 1
auto_learn_threshold_spam 5
..in my site config.
Any help gratefully received :)
Daz
-------------------------------------------------------
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
