Auto Whitelist

[infinityM]

Hey Guys,

I see PMG has an auto whitelist feature, but I don't actually see it being used somewhere.
Can someone help me understand this feature?

Also, Spam Auto learn... How does that work in PMG? I saw ham is auto learned, but I don't see spam auto learning.
I would assume a score threshold can be set, say 20, and anything above that is learned as spam for bayes?

SA score=22/5 time=3.941 bayes=undefined autolearn=spam autolearn_force=no hits=DKIM_INVALID(0.1),DKIM_SIGNED(0.1),FSL_BULK_SIG(0.001),HTML_FONT_LOW_CONTRAST(0.001),HTML_MESSAGE(0.001),KAM_INFOUSMEBIZ(0.75),KAM_VERY_BLACK_DBL(5),MIME_HTML_MOSTLY(0.1),MPART_ALT_DIFF(0.724),RAZOR2_CF_RANGE_51_100(2.43),RAZOR2_CHECK(1.729),RCVD_IN_PSBL(2.7),RCVD_IN_SBL_CSS(3.558),RDNS_NONE(1.274),SPF_HELO_NONE(0.001),SPF_PASS(-0.001),T_REMOTE_IMAGE(0.01),URIBL_BLACK(1.7),URIBL_DBL_SPAM(2.5)

Above is one of the lines for SA, as you can see it does not get a bayes result, Although it does seem to be learning spam & ham... But again, bayes undefined... So not sure if it's actually working...

Then on a last note, How can SA be taught manually?
Is there a way in Proxmox to set that up or handle that?

 

[Stoiko Ivanov]

I see PMG has an auto whitelist feature, but I don't actually see it being used somewhere.
Can someone help me understand this feature?

The feature basically enables the corresponding spamassassin feature:
https://cwiki.apache.org/confluence/display/SPAMASSASSIN/AutoWhitelist

Also, Spam Auto learn... How does that work in PMG? I saw ham is auto learned, but I don't see spam auto learning.
I would assume a score threshold can be set, say 20, and anything above that is learned as spam for bayes?

This particular mail got learned as spam (`autolearn=spam`)
https://cwiki.apache.org/confluence/display/SPAMASSASSIN/AutolearningNotWorking

However for bayes to start working it needs to learn quite a few mails (both ham and spam):

By default, there must be 200 ham messages and 200 spam messages learned before it will be used.

from https://cwiki.apache.org/confluence/display/SPAMASSASSIN/BayesNotWorking

Then on a last note, How can SA be taught manually?

you can use the sa-learn executable on mails, which you classified - but keep in mind that the mails need to remain unmodified (here also the SpamAssassin documentation is quite good)

I hope this helps!

 

[infinityM]

The feature basically enables the corresponding spamassassin feature:
https://cwiki.apache.org/confluence/display/SPAMASSASSIN/AutoWhitelist


This particular mail got learned as spam (`autolearn=spam`)
https://cwiki.apache.org/confluence/display/SPAMASSASSIN/AutolearningNotWorking

However for bayes to start working it needs to learn quite a few mails (both ham and spam):

from https://cwiki.apache.org/confluence/display/SPAMASSASSIN/BayesNotWorking


you can use the sa-learn executable on mails, which you classified - but keep in mind that the mails need to remain unmodified (here also the SpamAssassin documentation is quite good)

I hope this helps!

Would it not be practical for there to be a way to view previous emails, and teach SA based on that?
I know Mailscanner FE managed to include this quite effectively in their cPanel plugin which also uses SpamAssassin