Issue SpamAssassin not learning... ever

[ReneRoth]

Running Plesk for a few years now for my personal mail, and it just seems that the spam filter / SpamAssassin is not learning, at all.
I keep getting the exact same spam messages straight into my inbox over and over and over again, for months now, and always move them to the spam folder by hand. Yet Plesk never learns that those are, in fact, spam. The spam score of those does not increase at all.
How do I go about debugging this behaviour? The `autolearn` header flag stays at `no` for all mails, which seems to indicate that SpamAssassin is not learning from the messages I move by hand?

 

[Maarten.]

Lots of people seem to be confused by the "autolearn=no" statement in the default X-Spam-Status header:

AutolearningNotWorking - SPAMASSASSIN - Apache Software Foundation

cwiki.apache.org

 

[Maarten.]

I usually create a rule in /etc/mail/spamassassin/local.cf

WritingRules - SPAMASSASSIN - Apache Software Foundation

cwiki.apache.org

Or I create a filter in webmail in which I search for recurring words and move the mail to the spamfolder. This works really well and you don't have to be logged in to webmail.

 

[ReneRoth]

I usually create a rule in /etc/mail/spamassassin/local.cf

WritingRules - SPAMASSASSIN - Apache Software Foundation

cwiki.apache.org

Or I create a filter in webmail in which I search for recurring words and move the mail to the spamfolder. This works really well and you don't have to be logged in to webmail.

That absolutely defeats the point of SpamAssassin if I have to do that by hand - at this point I might as well switch to Google Mail and save myself a lot of headaches

 

[Kaspar]

That absolutely defeats the point of SpamAssassin if I have to do that by hand - at this point I might as well switch to Google Mail and save myself a lot of headaches

If you thought SpamAssassin has the same capabilities as Google mail in terms of automatic spam recognition you had/have (very) wrong expectations.

Neglecting the fact Google's spam AI is probably much more advanced. It also has millions of mailboxes to learn from. While you may just have a couple of mailboxes on your server?

There are some great Plesk extensions to improve spam filtering. SpamExperts (which can also be used without an extension) or Warden Anti-spam and Virus Protection for example. None of those are free. And perhaps are not as "good" as Google mail. But much better than SpamAssassin with it's default configuration.

 

[ReneRoth]

If you thought SpamAssassin has the same capabilities as Google mail in terms of automatic spam recognition you had/have (very) wrong expectations.

I did not think that for a moment, merely implied that I might be willing to trade off my privacy for better spam protection.
I am also aware Plesk is more interested in upselling, rather than, for example, keeping the SpamAssassin version bundled with their product up to date.
Thank you for your input.

To anyone else coming across this, I recommend:
- manually updating SpamAssassin (this alone makes a huge difference, plesk is around ten years behind)
- updating spamassassin rules from a big aggregator like mailbox.org
- installing the pyzor and razor addons
- adjusting the spam score based on IP geolocation
- training ham from the read mails based on a cronjob
- if you want to go the extra mile, periodically download spam mails and train those too

 

[Maarten.]

The "KAM Ruleset" is also a great help in fighting spam:

The McGrail Foundation - Projects

mcgrail.com

 

[Kaspar]

To anyone else coming across this, I recommend:
- manually updating SpamAssassin (this alone makes a huge difference, plesk is around ten years behind)
- updating spamassassin rules from a big aggregator like mailbox.org
- installing the pyzor and razor addons
- adjusting the spam score based on IP geolocation
- training ham from the read mails based on a cronjob
- if you want to go the extra mile, periodically download spam mails and train those too

Those are great suggestions. Thank you for sharing.

I did not think that for a moment, merely implied that I might be willing to trade off my privacy for better spam protection.
I am also aware Plesk is more interested in upselling, rather than, for example, keeping the SpamAssassin version bundled with their product up to date.
Thank you for your input.

With the exception of Windows, Plesk is merely utilizing the SpamAssassin version that comes with the OS. And yes, that is likely to be an older version. Again, Windows being the exception here, on which Plesk does keeps SA updated (sort of).