Whitelist

[hodel]

Is there a way to maintain a whitelist of email address? I have tried using the 'zmtrainsa' script to record ham for a certain email, but it is still being marked as spam.

Ray

[marcmac]

Which tests are catching it as spam? If you rt click the message and "show original" there will be a header with the score, and which tests contributed to that score.

[hodel]

My kill % is: 33
My Tag % is: 10

I'm not sure what those mean, but I believe I saw someplace in the forums that those values work well.

Here are a few test results from the mail headers:

X-Spam-Status: Yes, hits=4.702 tagged_above=-10 required=2 autolearn=no
tests=[BAYES_99=3.5, HTML_80_90=0.146, HTML_FONT_TINY=0.964,
HTML_TEXT_AFTER_BODY=0.061, HTML_TEXT_AFTER_HTML=0.031]
X-Spam-Level: ****
X-Spam-Flag: YES

------------------------------------------------------------------------

X-Spam-Status: Yes, hits=5.705 tagged_above=-10 required=2 autolearn=no
tests=[BAYES_99=3.5, DNS_FROM_RFC_POST=1.614, DNS_FROM_RFC_WHOIS=0.296,
HTML_80_90=0.146, HTML_FONT_BIG=0.142, MIME_BOUND_NEXTPART=0,
NO_REAL_NAME=0.007]
X-Spam-Level: *****
X-Spam-Flag: YES

------------------------------------------------------------------------

X-Spam-Status: Yes, hits=3.753 tagged_above=-10 required=2 autolearn=no
tests=[BAYES_99=3.5, HTML_10_20=0.246, MIME_BOUND_NEXTPART=0,
NO_REAL_NAME=0.007]
X-Spam-Level: ***
X-Spam-Flag: YES

[phoenix]

I would say those levels are too severe, have a look at this thread IMO, you should leave the settings on the lax side until you've got spamassassin trained a bit better as you're obviously picking up too many false-positives.

[hodel]

Okay. I'll make those suggested changes.

Thanks,

Ray

[hodel]

So the higher the kill and tag values the more spam that will get thru?

[phoenix]

Yes, that's true but you don't want to set it too low or you'll get the false positives that you have already been seeing. To some extent stopping spam is a bit of a juggling act between catching most of the spam and not having any of your 'real' mail classified as spam. Over time the training will catch the spam better and you could always add DSPAM, it's been mentioned in the forums and I'm using it.