Adding SpamAssassin report to email


#1

Hi,

I’ve noticed recently that SpamAssassin isn’t marking Spam as such. One spam email I’ve received recently has a score of -5.0.

I’d like to enable the ‘Spam-Report’ header for a while, so that I can see what rules are causing spam to be mis-marked. Anyone know how to do this in Symbiosis?

Thanks

Andy


#2

Hi Andy

Sorry, no direct answer but I’m also interested … I tried various incantations in

/srv/.spamassassin/user_prefs /etc/spamassassin/local.cf

# MC EDIT: temporarily add X-Spam-SA-Debug-001 header
# add_header all SA-Debug-001 _TESTSSCORES(,)_
#
add_header all SA-Debug-001 _SUMMARY_

# temp debug (less likely to be deleted)
# add_header all Checker-Version SpamAssassin _VERSION_ (_SUBVERSION_)

… restarted spamassassin (even exim4) but nothing showed up in subsequent test messages.

Fwiw, /var/log/mail.log shows scores and hits;

Jan  9 05:36:45 vm1 spamd[15093]: spamd: result: . -1 - BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,FREEMAIL_FROM,HTML_MESSAGE,SPF_PASS scantime=2.0,size=9575,user=admin,uid=1000,required_score=4.5,rhost=localhost.localdomain,raddr=127.0.0.1,rport=41485,mid=<CADLefieA_P6f=0sqAD03fk4yJB=Goa9bh4qH6WDXA-P7A7X5fQ@mail.gmail.com>,bayes=0.000000,autolearn=ham autolearn_force=no 
Jan 11 11:37:41 vm1 spamd[26898]: spamd: result: Y 5 - BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,HTML_IMAGE_RATIO_02,HTML_MESSAGE,HTML_MIME_NO_HTML_TAG,MIME_HTML_ONLY,PYZOR_CHECK,RDNS_NONE,URIBL_BLOCKED,URIBL_DBL_SPAM,URIBL_SBL,URIBL_SBL_A scantime=3.0,size=5391,user=admin,uid=1000,required_score=4.5,rhost=localhost.localdomain,raddr=127.0.0.1,rport=58019,mid=<81cf3697524e2f72e76fe1dde7bb99e7@tara.snxu.cn>,bayes=0.000000,autolearn=no autolearn_force=no
Jan 11 14:09:22 vm1 spamd[27216]: spamd: result: . -7 - BAD_CREDIT,BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,HTML_IMAGE_RATIO_08,HTML_MESSAGE,LOTS_OF_MONEY,RCVD_IN_MSPIKE_H4,RCVD_IN_MSPIKE_WL,RCVD_IN_RP_CERTIFIED,RCVD_IN_RP_SAFE,RP_MATCHES_RCVD,SPF_HELO_PASS,SPF_PASS,T_MONEY_PERCENT,URIBL_BLOCKED scantime=4.2,size=113895,user=admin,uid=1000,required_score=4.5,rhost=localhost.localdomain,raddr=127.0.0.1,rport=58802,mid=<1546665429.329136751484143753760.JavaMail.app@rbg41.atlis1>,bayes=0.000000,autolearn=ham autolearn_force=no
Jan 13 13:16:14 vm1 spamd[8855]: spamd: result: . 0 - BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,HTML_MESSAGE,RCVD_IN_MSPIKE_H3,RCVD_IN_MSPIKE_WL,RCVD_IN_SORBS_SPAM,RDNS_DYNAMIC,SPF_PASS,TVD_RCVD_IP,UNPARSEABLE_RELAY,URIBL_BLOCKED scantime=2.9,size=8755,user=admin,uid=1000,required_score=4.5,rhost=localhost.localdomain,raddr=127.0.0.1,rport=45216,mid=<c5002897a6d15155ad52c6f0a97022f9@async.facebook.com>,bayes=0.000000,autolearn=no autolearn_force=no

I don’t think I’ve seen anything other than BAYES_00, hence Spamassassin bayes issue, maybe but I’m still none the wiser. I’ve recently tried manual training;

… but that’s only with a few hundred messages and has had no apparent effect, yet.


I’m still keen on improving defences - it’s certainly improved with additional DNS RBL and I suspect spamassassin performance can be improved considerably - a local dns resolver (avoiding URIBL_BLOCKED) and even sieve might work wonders - but I’m wondering if, ultimately, it means using external email services. (Disabling ClamAV isn’t a good option in my case).