Teaching Mailcow how to deal with Ham/Spam

The good must be put in the dish, the bad you may eat if you wish.

Cinderella

Mailcow is a groupware solutions, that is mainly used for email messaging. With Mailcow, you can setup your own Docker-based Mail-Server + Addons.

Mailcow uses rspamd to filter out Spam Messages.
However, after some time, there is a need to fine-tune the Spam (Spam Messages)/Ham (“good” Messages) filtering.

There is a documented method to learn Spam from existing emails within a directory, but especially for non-technical users, that might be hard to understand.

So I updated this method a little bit:

• every user hast two folders rspamd/spam and rspamd/ham in their home directory.
• Every user can new just drop new spam messages into the spam and false spam messages into the ham folder.
• A cron Jobs runs every hour to parse the user directories for new files and updates the rspamd behaviour.

The script for SPAM learning looks like this (assumed that mailcow is installed in /opt/mailcow-dockerized):

#!/bin/bash

cd /opt/mailcow-dockerized
for u in $(ls /home); do
    ""mv /home/$u/rspamd/spam/* ./data/rspamd/spam/""
done
for file in ""./data/rspamd/spam/*""; do 
    docker exec -i $(docker-compose ps -q rspamd-mailcow) rspamc learn_spam < "$file"
done

""rm -Rf ./data/rspamd/spam/*""

There is also a similar script for HAM learning:

#!/bin/bash

cd /opt/mailcow-dockerized
for u in $(ls /home); do
    ""mv /home/$u/rspamd/ham/* ./data/rspamd/ham/""
done
for file in ""./data/rspamd/ham/*""; do 
    docker exec -i $(docker-compose ps -q rspamd-mailcow) rspamc learn_ham < "$file"
done
""rm -Rf ./data/rspamd/ham/*""

Both scripts will produce some output, so a good way of running it via cron, is to pipe the output into a log file.