Well. Three senders of that "bad" domain made it into the users'
(auto)whitelist.
I'm not quite sure why. But now I just remove those hoping they won't appear
there again.
Thanks for helping!
--
View this message in context:
http://www.nabble.com/Spamassassin-doesn%27t-learn---debug-outputs-t
>>sa-learn --dump magic
>>if both nham, nspam is over 200 then show me
>>
>>spamassassin 2>&1 -D -t /tmp/log
>>and maybe olso
>>spamassassin 2>&1 -D --lint >/tmp/lint
>>
>>post log and lint file somewhere
>
>
First: The dump-data:
0.000 0 3 0 non-token data: bayes
Benny Pedersen wrote:
>
> seems ok :-)
>
> /usr/bin/fetchmail -a -s -n --folder assassin/spam -m '/usr/bin/sa-learn
> 2>&1
> -D --spam' >>/var/log/assassinspam.log
>
> i hope its this, not tested
>
>
That helped (wrong position for 2>&1 but sometimes I surprisingly manage to
think for mysel
Hi there. It seems my Spamassassin does not learn very well...
I use the following statements to learn spam/ham:
/usr/bin/fetchmail -a -s -n --folder assassin/spam -m '/usr/bin/sa-learn -D
--spam' >>/var/log/assassinspam.log
/usr/bin/fetchmail -a -s -n --folder assassin/ham -m '/usr/bin/sa-learn
