Mailing List Archive

Martin Lyberg wrote:

>I have the following rule in my local.cf:
>
>whitelist_to *@incubator.apache.org
>
>Sometimes mail from this mailinglist is getting marked as spam, if i check
>the header i can see that the adjustment for the above rule is -6.0 points.
>Can i increase this somehow?
>
>
Probably, but are there high-scoring rules that are causing them to be
marked spam? Whatever adjustment you add needs to more than offset that
amount. What scores are they getting?

Considering the amount of "spam-content" in this group, and not wanting
to confuse bayes and AWL, I simply wrote a procmail rule to bypass SA
scoring of this group.

That said, I'm not a rules wiz, but I think:

header LOCAL_SALIST ListID =~ /users@spamassassin/
describe LOCAL_SALIST spamassassin discussion list
score LOCAL_SALIST -6.0

would do what you want.

- Bob

martin.lyberg@idkommunikation.com said:
> I have the following rule in my local.cf:
> whitelist_to *@incubator.apache.org
> Sometimes mail from this mailinglist is getting marked as spam, if i check
> the header i can see that the adjustment for the above rule is -6.0 points.
> Can i increase this somehow?

Try:

all_spam_to spamassassin-users@incubator.apache.org

Owen




--
Via Net.Works UK Ltd
Local Touch Global Reach
Owen McShane Systems Administrator
http://www.vianetworks.co.uk Tel +44 (0)1925 484444

Owen McShane wrote:

>all_spam_to spamassassin-users@incubator.apache.org
>
>
Hmm. The manpage mentions "whitelisting" of recipients with more_spam_to
and all_spam_to. These, and whitelist_to, are mutually exclusive, right?
Or are they cumulative? Can I use differing degrees of granularity for
each (i.e. whitelist_to *@domain, plus more_spam_to user@domain) (I
realize I can test, but I'm after the authoritative -- "it'll be this
way forever" -- answer).

After my recent experiences with white/blacklists impacting user AWLs,
I'm trying to do everything through modest scoring nudget rather than
white/blacklist big-point whacks.

- Bob

: From: Bob George [mailto:mailings02@ttlexceeded.com]

: Probably, but are there high-scoring rules that are causing
: them to be
: marked spam? Whatever adjustment you add needs to more than
: offset that
: amount. What scores are they getting?

It depends, most of the mails get through, but sometimes a adjustment of
-6.0 (SA default) points isn't enough to let it through.

:
: Considering the amount of "spam-content" in this group, and
: not wanting
: to confuse bayes and AWL, I simply wrote a procmail rule to bypass SA
: scoring of this group.
:
: That said, I'm not a rules wiz, but I think:
:
: header LOCAL_SALIST ListID =~ /users@spamassassin/
: describe LOCAL_SALIST spamassassin discussion list
: score LOCAL_SALIST -6.0
:
: would do what you want.

Should i place this rule in my local.cf?

Thank you!

: From: Owen McShane [mailto:omcshane@vianetworks.co.uk]
:
: Try:
:
: all_spam_to spamassassin-users@incubator.apache.org

Thank you, that seemed to solve my issue.

From: "Bob George" <mailings02@ttlexceeded.com>

> Martin Lyberg wrote:
>
> >I have the following rule in my local.cf:
> >
> >whitelist_to *@incubator.apache.org
> >
> >Sometimes mail from this mailinglist is getting marked as spam, if i
check
> >the header i can see that the adjustment for the above rule is -6.0
points.
> >Can i increase this somehow?
> >
> >
> Probably, but are there high-scoring rules that are causing them to be
> marked spam? Whatever adjustment you add needs to more than offset that
> amount. What scores are they getting?
>
> Considering the amount of "spam-content" in this group, and not wanting
> to confuse bayes and AWL, I simply wrote a procmail rule to bypass SA
> scoring of this group.
>
> That said, I'm not a rules wiz, but I think:
>
> header LOCAL_SALIST ListID =~ /users@spamassassin/
> describe LOCAL_SALIST spamassassin discussion list
> score LOCAL_SALIST -6.0
>
> would do what you want.

We've found that to do it all the way you need something more like:
--8<--
# Find msgs to the stupid SA mailing list. grrr...

header __SA_TO To =~ /\@incubator\.apache\.org/i
header __SA_FROM From =~ /\@incubator\.apache\.org/i
header __SA_CC Cc =~ /\@incubator\.apache\.org/i
header __SA__ID List-Id =~ /spamassassin\.apache\.org/i
meta SA_LIST_MAIL ((__SA_TO || __SA_FROM || __SA_CC ||
__SA_ID)>0)
describe SA_LIST_MAIL Mail from Spamassassin mailing list
score SA_LIST_MAIL -50

header SA_TO To =~ /\@incubator\.apache\.org/i
header SA_FROM From =~ /\@incubator\.apache\.org/i
header SA_CC Cc =~ /\@incubator\.apache\.org/i
header SA__ID List-Id =~ /spamassassin\.apache\.org/i
score SA_TO -0.01
score SA_FROM -0.02
score SA_CC -0.03
score SA_ID -0.04
describe SA_TO To SA list
describe SA_FROM From SA list
describe SA_CC Cc SA list
describe SA_ID SA list ID
--8<--

Note that you had "List-ID" spelled incorrectly above.
{^_-}

"jdow" <jdow@earthlink.net> wrote:
> We've found that to do it all the way you need something more like:
> --8<--
> # Find msgs to the stupid SA mailing list. grrr...
>
> header __SA_TO To =~ /\@incubator\.apache\.org/i
> [...]

I am definitely leaning towards using scored rules (like your samples) rather
than black/whitelists. I may give my awk/perl coding a run to see if I can
convert some of the blacklist tables I've collected. I decided to just avoid
the issue, and not score spam-related lists lest I go nuts constantly tweaking
'em!

Do you use AWL or autolearn? I've found that BIG score swings (greater than
spam_threshold) can cause long-lasting AWL issues. I'm shying away from big +/-
scores as a result.

> [...]
> Note that you had "List-ID" spelled incorrectly above.

It's this funny accent I've developed since moving to the northeast. I'm
dropping hyphens! (good catch, thanks)

- Bob

From: "Bob George" <mailings02@ttlexceeded.com>

> "jdow" <jdow@earthlink.net> wrote:

> Do you use AWL or autolearn? I've found that BIG score swings (greater
than
> spam_threshold) can cause long-lasting AWL issues. I'm shying away from
big +/-
> scores as a result.

No no. No no no. No nononono! I don't. And the only reason I don't is
that initially the comments that came out with the spam messages somewhat
confused me, "Learn to auto WHITE list spam, will you?" I turned it off
with great haste. Later the Wiki corrected my misconception. However, the
episode forced me to think about the implications of autowhitelist; and I
decided to leave it off along with autolearn. They are a double eended
sword. They cut the target and the holder of the sword about equally.

Since I made that decision mostly on intuition I have decided I very
probably made a good decision, especially in as much as I use manual
white listing and blacklisting quite often. (I'm about to pull paypal
and ebay off the white list and simply write rules with sane values
and more imaginative complexitivity for them.)

> > [...]
> > Note that you had "List-ID" spelled incorrectly above.
>
> It's this funny accent I've developed since moving to the northeast. I'm
> dropping hyphens! (good catch, thanks)

De nada.
{^_-} (Since moving from looking South on one Ontario that is well North
and East of the one I now look South upon I've picked up an even
funnier accent.)