This is the Docker stack to set up a fully functional AgentJ antispam system at a glance.

The AgentJ Docker stack is composed of the following services:

• db: a MariaDB instance, it keeps track of the mail headers and other information needed to manage the e-mails life cycle (sender, recipient, amavis id, ...)
• app: a Web UI that allows you to add and manage your domains and associated users as well as managing the incoming e-mails (block, release, white/black lists)
• web: a Nginx based reverse proxy that serves the app
• stmp: a Postfix instance that will receive the e-mails and forward them to the amavis container (Amavis/ClamAV/Spamassassin service)
• amavis: a container running Amavis/Spamassassin and ClamAV services
• redis: a Redis instance used as cache for the app
• logspout + syslogng: a Syslog-NG instance that will collect and centralize logs from the other containers
• relay: an other Postfix instance, needed to avoid loops when forwarding the released or white-listed e-mails to their receipient(s)

All you have to do is to clone the repository

git clone https://github.com/Probesys/agentj-docker.git

then, cd to the cloned repository to configure a few variables:

cd agentj-docker

The following runtime variables must be configured:

The AgentJ antispam stack has its own Docker bridge br-agentj and IPv4 subnet which defaults to 172.42.42.0/24 (configurable, see variables table above).

It is not recommended to launch the stack as root. We recommend you to create a dedicated docker user (make sure it belongs to the docker group). After you have set the above variables, you can start the stack with following commands:

docker-compose up -d

The Web UI will be available at http://hostname:8090. The default login is admin and the default password is lutte antispam (yes, this is a space between the two words).

When started, the AgentJ stack will create the following volumes:

• public: the app sources files
• db : the MariaDB databases files
• redisdb: Redis DB dumps
• clamdb: the ClamAV virus signatures database, updated 4 times a day
• quarantine: the quarantined e-mails
• logs: the log files from all containers, centralized by the syslogng container

In order to upgrade the AgentJ dockerized stack, you must stop the running containers, pull (or build locally if you prefer) the updated images, remove the public volume and start the updated stack:

docker-compose down
# To upgrade only the webapp, delete and update its image
docker image rm $(docker image ls -q -f reference=*/*agentj_app)
# To upgrade all the stack, delete and update all images
docker image rm $(docker image ls -q -f reference=*/*agentj_*)
docker volume rm agentj-docker_public
docker-compose up -d

This work is made available under the GNU Affero General Public License v3.0.

AgentJ is a Probesys project.