mirror of
https://github.com/zokradonh/kopano-docker
synced 2025-06-22 23:36:14 +00:00
32 lines
1.7 KiB
Markdown
32 lines
1.7 KiB
Markdown
# kopano-spamd extras for kopano-docker
|
|
|
|
This directory contains a compose file including optional containers to enable learning spam to be used with the [SpamAssassin](https://spamassassin.apache.org/) bayes filter.
|
|
|
|
## How to use this compose file?
|
|
|
|
1. Add the `spamd-extras.yml` to the `COMPOSE_FILE` variable in your `.env` file.
|
|
|
|
Example:
|
|
|
|
```bash
|
|
COMPOSE_FILE=docker-compose.yml:docker-compose.ports.yml:spamd-extras/spamd-extras.yml
|
|
```
|
|
|
|
2. Run `docker-compose up -d`.
|
|
|
|
|
|
## kopano-spamd
|
|
|
|
After startup there will be a new service `kopano-spamd` which will persist mails that are moved to `Junk` to a folder named `spam` inside the `kopanospamd` volume.
|
|
Likewise mails that are moved from the `Junk` back to `Inbox` are persisted in `ham`. Both folders indicate mails that should be trained as either being ham or spam.
|
|
|
|
The `kopano-scheduler` container is extended to run the training inside the `mail` docker container at about 4am with training results being picked up by SpamAssassin
|
|
automatically. You can check the docker logs of the scheduler for the results of each run.
|
|
|
|
While already trained files can be deleted immediately after each training run, there is no cleanup provided here. The `kopanospamd` volume will therefore grow over time.
|
|
|
|
For the bayes filter to start working you will need to train at least 200 mails of each ham and spam. To create a set of initial data you can use the Kopano WebApp
|
|
by selecting mails and using the `Export as` function to create a zip file of those mails and put them into the appropriate folder.
|
|
|
|
Read more about how to create effective training data here: https://spamassassin.apache.org/full/3.4.x/doc/sa-learn.html#EFFECTIVE-TRAINING
|