Welcome to this technology demonstrator where we setup a 3 node HiveMQ cluster along with a HA proxy that acts as a load balancer and single IP entrypoint to this cluster.

Start by cloning this repo to your docker and docker compose enabled platform and CD into it. Then set the HiveMQ version that you want to deploy and build the local HiveMQ image. This image will include the HiveMQ config.xlm file that enables the cluster:

<cluster>
    <enabled>true</enabled>
    <transport>
        <tcp>
            <!-- replace this IP with the IP address of your interface -->
        </tcp>
    </transport>
    <discovery>
        <broadcast>
            <!-- replace this IP with the broacast IP address of your subnet -->
        </broadcast>
    </discovery>
</cluster>

Please refer to the following commands:

git clone https://github.com/hivemq/hivemq-HAproxy.git
cd hivemq-HAproxy
export HIVEMQ_VERSION=4.9.0
sudo chmod +x build.sh
./build.sh
docker-compose up

If you run a docker ps the output will show 3 HiveMQ nodes running along with one HAproxy instance:

Checkout: http://0.0.0.0:8404/stats Are all nodes detected by the loadbalancer ?

Checkout: http://0.0.0.0:8080 Are all nodes running and formed a cluster as shown in the HiveMQ Control center (default password: admin/hivemq) ?

HiveMQ CLI tools https://www.hivemq.com/blog/mqtt-cli/

Test with : Mqtt sub -h 0.0.0.0 -t "testtopic" in a seperate CLI window and Mqtt pub -h 0.0.0.0 -m "test" -t "testtopic"

You can simulate malfunction of one of the nodes by simply killing it: docker kill a21ec7267a6d

Now the node has been gone wich is shows in HiveMQ CC and in the HAproxy stats screen:

You can restore the original 3 node configuration simply by re-running docker-compose up.

Please clean, up afterwards with docker-compose down.

For contact information please reach out to [email protected]

extend this demo to include a roling version upgrade.