Spam score - a sum of all listed tests or not?

Damian amavis at arcsin.de
Tue Oct 13 11:21:44 CEST 2020


It seems that you are using Rspamd as an additional spam scanner. If so,
I see some possibilities.

1) Your @spam_scanners score_factor for Rspamd is something peculiar.
2) The score reported back by Rspamd is not the sum of reported symbol
scores.
3) The amavis-internal RspamdClient should incorporate the score_factor
into the values of reported tests.

With log_level 5 you could see the raw json rspamd response. It would be
helpful to check 2).

> Here I have a real-world example - this is the X-Spam-Satusheader of a
> received spam message (I added newlines for clarity):
>
> X-Spam-Status: No, score=-0.499 tagged_above=-999 required=4 tests=[
> DKIMWL_WL_MED=-0.001,
> DKIM_SIGNED=0.1,
> DKIM_VALID=-0.1,
> DKIM_VALID_AU=-0.1,
> HTML_MESSAGE=0.001,
> MAILING_LIST_MULTI=-1,
> RCVD_IN_MSPIKE_H4=-0.01,
> RCVD_IN_MSPIKE_WL=-0.01,
> SPF_HELO_NONE=0.001,
> SPF_PASS=-0.001,
> T_KAM_HTML_FONT_INVALID=0.01,
> RCVD_TLS_ALL=0,
> RWL_MAILSPIKE_VERYGOOD=0,
> HAS_LIST_UNSUB=-0.01,
> R_DKIM_ALLOW=0,
> URIBL_GREY=1.5,
> MANY_INVISIBLE_PARTS=0.3,
> DMARC_POLICY_ALLOW=0,
> BAD_REP_POLICIES=0.1,
> REPLYTO_DOM_NEQ_FROM_DOM=0,
> PREVIOUSLY_DELIVERED=0,
> PRECEDENCE_BULK=0,
> DKIM_TRACE=0,
> MIME_TRACE=0,
> MIME_BASE64_TEXT=0.1,
> HAS_REPLYTO=0,
> RCVD_COUNT_ONE=0,
> ASN=0,
> MIME_GOOD=-0.1,
> TO_DN_EQ_ADDR_ALL=0,
> R_SPF_ALLOW=0,
> FREEMAIL_REPLYTO=0,
> FORGED_SENDER=0.3,
> TO_MATCH_ENVRCPT_ALL=0,
> ONCE_RECEIVED=0.1,
> ARC_NA=0,
> RCPT_COUNT_ONE=0,
> FROM_NEQ_ENVFROM=0,
> FROM_HAS_DN=0
> ] |autolearn=ham autolearn_force=no|
>
> So - total score is -0.499, but if I manually sum up all the tests, it
> comes out to 1.18.
>
> Is this something that normally occurs in Amavis or is this a distro
> (CentOS 7) specific problem?
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.amavis.org/pipermail/amavis-users/attachments/20201013/ff286a19/attachment.htm>


More information about the amavis-users mailing list