You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by shane mullins <ts...@wise.k12.va.us> on 2004/12/13 16:47:03 UTC
SA-LEARN feature
The sa-learn feature is great. And, I know, this has been brought up
before. But, could someone please tell me how to get the spam/ham
messages back to our spam filters? Our spam filters are smtp gateways
only. We use OpenBSD 3.5, Spamassassin 3.0.1, dcc, razor and amavisd.
Shane
Re: SA-LEARN feature
Posted by Thomas Arend <ml...@arend-whv.info>.
Am Montag, 13. Dezember 2004 17:13 schrieb shane mullins:
> Thanks Ronan,
>
> When I run Spamassasin -D --lint, I get the following:
>
> debug: bayes: Not available for scanning, only 0 spam(s) in Bayes DB <
> 200
>
> If I could get my >200 spam/ham's in there, then I could start using
> Bayes. SA is working great, but if I could get Bayes working, it would
> be even better.
>
> Shane
>
>
You can reduce the required amount of spam / ham in user_prefs.cf
or /etc/mail/spammassain/locale.cf
bayes_min_ham_num 100
#(Default: 200)
bayes_min_spam_num 100
#(Default: 200)
Thomas
[..]
--
icq:133073900
aim:tawhv
Re: SA-LEARN feature
Posted by shane mullins <ts...@wise.k12.va.us>.
Thanks Ronan,
When I run Spamassasin -D --lint, I get the following:
debug: bayes: Not available for scanning, only 0 spam(s) in Bayes DB <
200
If I could get my >200 spam/ham's in there, then I could start using
Bayes. SA is working great, but if I could get Bayes working, it would
be even better.
Shane
----- Original Message -----
From: "Ronan" <r....@qub.ac.uk>
To: "shane mullins" <ts...@wise.k12.va.us>
Sent: Monday, December 13, 2004 10:53 AM
Subject: Re: SA-LEARN feature
> we run the same config smtp exim w/ exiscan ACL filters virus' first
> then tags the message with SA headers and jsut forwards it on to the
> approriate server.
> using autolearn (which shuold be enabled by default) and by running
200
> messages spam / ham through sa-learn on the filters and have your
> baysian DB set up properly ( permissions etc) autolearn will kick in
and
> you can start building up your totals...
>
> may be firrenet for you but i get the following from /var/log/syslog
>
> ^C# grep autolearn /var/log/syslog
> Dec 13 03:37:11 elisha spamd[11159]: result: Y 33 -
>
BAYES_99,CLICK_BELOW_CAPS,DATE_IN_PAST_96_XX,FORGED_IMS_HTML,FORGED_IMS_
TAGS,FORGED_MUA_IMS,FROM_HAS_MIXED_NUMS3,HTML_50_60,HTML_MESSAGE,HTML_MI
ME_NO_HTML_TAG,MIME_HTML_ONLY,RCVD_AM_PM,RCVD_HELO_IP_MISMATCH,RCVD_IN_B
L_SPAMCOP_NET,RCVD_IN_DSBL,RCVD_IN_NJABL_DUL,RCVD_IN_NJABL_PROXY,RCVD_IN
_SORBS_DUL,RCVD_IN_SORBS_HTTP,RCVD_IN_SORBS_MISC,RCVD_IN_XBL,RCVD_NUMERI
C_HELO,SARE_OEM_PRODS_1,SARE_OEM_PRODS_2,SARE_OEM_PRODS_FEW,SARE_OEM_SOF
T_IS,SARE_PRODUCTS_02,SARE_PRODUCTS_03
> scantime=7.7,size=2450,mid=(unknown),bayes=1,autolearn=spam
> Dec 13 03:37:21 elisha spamd[6049]: result: Y 33 -
>
BAYES_99,CLICK_BELOW_CAPS,DATE_IN_PAST_96_XX,FORGED_IMS_HTML,FORGED_IMS_
TAGS,FORGED_MUA_IMS,FROM_HAS_MIXED_NUMS3,HTML_50_60,HTML_MESSAGE,HTML_MI
ME_NO_HTML_TAG,MIME_HTML_ONLY,RCVD_AM_PM,RCVD_HELO_IP_MISMATCH,RCVD_IN_B
L_SPAMCOP_NET,RCVD_IN_DSBL,RCVD_IN_NJABL_DUL,RCVD_IN_NJABL_PROXY,RCVD_IN
_SORBS_DUL,RCVD_IN_SORBS_HTTP,RCVD_IN_SORBS_MISC,RCVD_IN_XBL,RCVD_NUMERI
C_HELO,SARE_OEM_PRODS_1,SARE_OEM_PRODS_2,SARE_OEM_PRODS_FEW,SARE_OEM_SOF
T_IS,SARE_PRODUCTS_02,SARE_PRODUCTS_03
> scantime=8.6,size=2449,mid=(unknown),bayes=1,autolearn=spam
> Dec 13 03:37:27 elisha spamd[14415]: result: Y 37 -
>
BAYES_99,CLICK_BELOW_CAPS,DATE_IN_PAST_96_XX,FORGED_IMS_HTML,FORGED_IMS_
TAGS,FORGED_MUA_IMS,FROM_HAS_MIXED_NUMS3,HTML_50_60,HTML_MESSAGE,HTML_MI
ME_NO_HTML_TAG,MIME_HTML_ONLY,RCVD_AM_PM,RCVD_HELO_IP_MISMATCH,RCVD_IN_B
L_SPAMCOP_NET,RCVD_IN_DSBL,RCVD_IN_NJABL_DUL,RCVD_IN_NJABL_PROXY,RCVD_IN
_SORBS_DUL,RCVD_IN_SORBS_HTTP,RCVD_IN_SORBS_MISC,RCVD_IN_XBL,RCVD_NUMERI
C_HELO,SARE_OEM_PRODS_1,SARE_OEM_PRODS_2,SARE_OEM_PRODS_FEW,SARE_OEM_SOF
T_IS,SARE_PRODUCTS_02,SARE_PRODUCTS_03,URIBL_SBL
> scantime=4.6,size=2452,mid=(unknown),bayes=1,autolearn=spam
> ==snip==
>
> One thing to note though the DB's will be diffrerent as they're on
> seperate machines. So eg your mx preferenece 1 will probably build up
a
> more comprehensive DB quicker that your others...
>
> ronan
>
> shane mullins wrote:
> > The sa-learn feature is great. And, I know, this has been brought
up
> > before. But, could someone please tell me how to get the spam/ham
> > messages back to our spam filters? Our spam filters are smtp
gateways
> > only. We use OpenBSD 3.5, Spamassassin 3.0.1, dcc, razor and
amavisd.
> >
> > Shane
> >
> >
>
> --
> Regards
>
> Ronan McGlue
> ==============
> Analyst/Programmer
> Information Services
> Queens University Belfast
> BT7 1NN