On 02/04/19, Larry Rosenman via Exim-users (exim-users@???) wrote:
> On 04/02/2019 2:20 am, Rory Campbell-Lange via Exim-users wrote:
> > On 01/04/19, Larry Rosenman (ler@???) wrote:
> > > On 04/01/2019 2:01 pm, Rory Campbell-Lange via Exim-users wrote:
> > > > It's some years since I've spent time tweaking my exim setup to receive
> > > > spam. I've forgotten any skills I might once have had in this area.
> > > >
> > > > I've gotten sick of getting 30+ spam emails a day and need to do
> > > > something about it! I'd be grateful for some pointers to the
> > > > state-of-the-art setup.
> >
> > Glancing at the docs under chapter 35, I guess my local users are
> > "authenticated" due to our use of cram_md5. I'm giving your rules a go!
> >
> > I wonder also if my /etc/spamassasin/local.cf is right
> >
> > required_score 3.0
> > score RP_MATCHES_RCVD -0.01
> > bayes_auto_learn 0
> > ifplugin Mail::SpamAssassin::Plugin::Shortcircuit
> > endif # Mail::SpamAssassin::Plugin::Shortcircuit
> >
> > Required score seems quite a bit lower than 70 in Exim.
> >
> > Thanks again
> > Rory
> Please do *NOT* use the X-LERCTR header (that's my domain)....
Sorry about that Larry -- that was a test using your rules.
> Also, to get a nice compact report, I have smmsp's user_prefs set to:
> ⌂68% [root@???:~] # cat user_prefs
> clear_report_template
> report SpamScore (_SCORE_/_REQD_) _TESTSSCORES(,)_
> ⌂68% [root@???:~] #
>
> (smmsp is the user exim calls spamd as).
>
> note also that $spam_score_int is the score * 10.
Thank you. My scores are improving
However I'm still getting all BAYES_OO scores as -1.9. I assume this is
because I've not yet trained sa-learn with enough spam. (I've trained it
with several thousand spam messages).
