Anything Forums

By default spamassassin is not really strict, especially because it randomly gives -2 points using the rule ALL_TRUSTED.

Here's how you fix this.

Type

vim /usr/share/spamassassin/50_scores.cf

(note: paths may be different on your system, you may need to use the locate command to find out where the .cf files are)

in vim type

/ALL_TRUSTED

You will find a line like this:

score ALL_TRUSTED -2.400 -2.820 -2.867 -3.300

Change the numbers to 0. The rule will still show up randomly, but at least it won't be removing points. I had a fix before to remove it completely but forget what I did.

Archived topic from Iceteks, old topic ID:3943, old post ID:32256

See, now that's where I'm glad my system uses SpamAssasin as only a small part of the overall filtering. It also uses a Challenge Response system and Bayesian Filtering (For those that don't know, Bayesian filters basically are like a learning filtering system. They look for characteristics of spam in the content, headers, word-pairs etc), so there's really no ALL_TRUSTED rule for me that makes a difference

Archived topic from Iceteks, old topic ID:3943, old post ID:32306

Yeah I have bayes, but if it matches it only get's like 1 point. I need to figure how to tweak that too. I must of learned it off a couple thousand spam emails too, though much less ham (non spam) emails, maybe like 200.

Though with my new filtering system it will be very easy to add filters, and modify them. I just have to do a search query, find the filter that did not work properly (too little, or too many points) and modify it.

I've been wanting to use RBLs too, just no time to set it up.

Archived topic from Iceteks, old topic ID:3943, old post ID:32308