Page 1 of 1

How to teach efa a spam mail

Posted: 04 Jul 2020 13:04
by gobris
Hello
There is an email that keeps coming from different email providers, with moslty similar topics.
I tried to blacklist servers a few times but , that works only for a few days.. Then the change the email server and go on..

So, when I click on (i) on webpage.. there is no tab of sa-learn under the email (starangely it only apperars on emails that are already spam.)
2020-07-04_16-02-12.png
2020-07-04_16-02-12.png (21.42 KiB) Viewed 3669 times
So is there a way to select this email for spam and run sa-learn or anything similar to block future emails?

Best Regards

Re: How to teach efa a spam mail

Posted: 04 Jul 2020 15:04
by shawniverson
You would probably get better results if you fixed your URIBL failures.

Also, if you aren't storing non-spam, you can't sa-learn, because there is no message stored.

Re: How to teach efa a spam mail

Posted: 06 Jul 2020 12:38
by pdwalker
Your bayes classifier is not running. As shawn says, or you've disabled bayes, or not enough mail has come in for bayes to do its job.

Re: How to teach efa a spam mail

Posted: 07 Jul 2020 09:10
by gobris
Ok.. I enabled unbound now uridb and dcc is working fine.. noticed solved another problem while looking for other.
So where can I enable to store non-spam emails for sa-learn

Re: How to teach efa a spam mail

Posted: 07 Jul 2020 11:10
by pdwalker
edit /etc/MailScanner/MailScanner.conf

Look for "Spam Actions =". Add "store" to the line if it is not there already.

Look for "High Scoring Spam Actions =". Add "store" to the line if it is not there already.

Look for "Non Spam Actions =". Add "store" to the line if it is not there already.

Now, a copy of all mail will also be stored on your EFA box. Keep your box secure.


Also, check your bayes database, under the EFA mailwatch ui, "Tools and Links", "SpamAssassin Bayes Database Info". Hopefully you'll see something like this:
Screen Shot 2020-07-07 at 19.12.png
Screen Shot 2020-07-07 at 19.12.png (54.75 KiB) Viewed 3615 times