Set up Kafka extension using HiveMQ operator

Install Kafka using helm

1. Create a namespace for Kafka and switch the context to it:

   kubectl create namespace kafka;  

   kubectl config set-context --current --namespace=kafka

2. Add the repository for the Kafka Helm chart to your package manager.

   helm repo add bitnami https://charts.bitnami.com/bitnami

   helm repo update

3. Deploy the Kafka server using the Helm chart.

   1. Additional step with EKS cluster:
      You will need to configure the Amazon EBS CSI driver. You can find detailed steps How to configure Amazon EBS CSI driver for working PersistentVolumes in EKS

   2. The below command deploys Kafka with 2 brokers (replicas).

      helm upgrade --install kafka bitnami/kafka --namespace=kafka --set replicaCount=2 

4. Please notice the output of the command above, it provides critical data that is used for the next steps

   1. Consumers can access Kafka via port 9092 on the following DNS name from within your cluster: kafka.kafka.svc.cluster.local

   2. The CLIENT listener for Kafka client connections from within your cluster has been configured with the following security settings: SASL authentication

   3. To connect a client to your Kafka:

      1. username="user1"

      2. To get the password execute the command below:(skip % at the end)

         kubectl get secret kafka-user-passwords --namespace kafka \
           -o jsonpath='{.data.client-passwords}' | base64 -d | cut -d , -f 1;

Configuring the Kafka Extension

Setting up the Kafka license as a ConfigMap

1. HiveMQ Enterprise Extension For Kafka requires a separate license file, e.g. kafka-license.elic, in the $HIVEMQ_HOME/license directory. To add the kafka-license.elic along with the hivemq-license.lic, create a new configmap hivemq-license including all desired license files:

   kubectl create configmap hivemq-license --namespace=hivemq \
     --from-file hivemq-license.lic \
     --from-file kafka-license.elic

2. Edit the values.yaml file of the hivemq-operator, section hivemq.configMaps. Update this:

     configMaps: []
     # ConfigMaps to mount to the HiveMQ pods. These can be mounted to existing directories without shadowing the folder contents as well.
     #- name: hivemq-license
     #  path: /opt/hivemq/license

   To this:

     configMaps: 
       - name: hivemq-license
         path: /opt/hivemq/license

   This will mount the content of the configMap hivemq-license to the directory /opt/hivemq/license of the hivemq-broker pods.

Configuring the extension

1. Prepare a simple configuration file for kafka-extension as in the example below.

   • this example configuration will map all incoming MQTT publish packets to the topic “test” in Kafka; and will map the topic “test” in Kafka to the topic “test” in the HiveMQ broker

   • Use your password in <password>here_is_your_password</password>, that you successfully retrieved with this command a few steps ago:

     kubectl get secret kafka-user-passwords --namespace kafka \
       -o jsonpath='{.data.client-passwords}' | base64 -d | cut -d , -f 1;

   • Here is the file:

<?xml version="1.0" encoding="UTF-8" ?>
<kafka-configuration xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
                     xsi:noNamespaceSchemaLocation="config.xsd">
    <kafka-clusters>
        <kafka-cluster>
            <id>cluster01</id>
            <bootstrap-servers>kafka.kafka.svc.cluster.local:9092</bootstrap-servers>
            <authentication>
                <scram-sha256>
                    <username>user1</username>
                    <password>here_is_your_password</password>
                </scram-sha256>
            </authentication>
        </kafka-cluster>
    </kafka-clusters>

    <mqtt-to-kafka-mappings>
        <mqtt-to-kafka-mapping>
            <id>mapping01</id>
            <cluster-id>cluster01</cluster-id>
            <mqtt-topic-filters>
                <mqtt-topic-filter>#</mqtt-topic-filter>
            </mqtt-topic-filters>
            <kafka-topic>test</kafka-topic>
        </mqtt-to-kafka-mapping>
    </mqtt-to-kafka-mappings>

    <kafka-to-mqtt-mappings>
        <kafka-to-mqtt-mapping>
            <id>mapping02</id>
            <cluster-id>cluster01</cluster-id>
            <kafka-topics>
                <kafka-topic>test</kafka-topic>
                <kafka-topic-pattern>test-(.)*</kafka-topic-pattern>
            </kafka-topics>
        </kafka-to-mqtt-mapping>
    </kafka-to-mqtt-mappings>
</kafka-configuration>

2. Create a new configMap kafka-config from the config.xml:

kubectl create configmap kafka-config --from-file config.xml --namespace hivemq

3. Edit values.yaml file of hivemq-operator and update section hivemq.extensions, having name: hivemq-kafka-extension
   Update this:

     extensions:
       - name: hivemq-kafka-extension
         extensionUri: preinstalled
         enabled: false

   To this:

   extensions:
       - name: hivemq-kafka-extension
         extensionUri: preinstalled
         enabled: true
         configMap: kafka-config
         initialization: |
           # Fixes the location of the config.xml file
           [[ ! -f conf/config.xml ]] &&
           [[ -f /conf-override/extensions/hivemq-kafka-extension/config.xml ]] &&
           ln -s /conf-override/extensions/hivemq-kafka-extension/config.xml conf/config.xml

4. Re-deploy hivemq-operator with updated values.yaml

   helm upgrade hivemq --install hivemq/hivemq-operator --values values.yaml --namespace hivemq

   – for ease of use we switch namespace back to hivemq kubectl config set-context --current --namespace=hivemq

Check if the license and configuration is applied correctly

If everything is correct,

1. The HiveMQ log contains info about using the correct license:

kubectl logs deployment/hivemq | grep 'Using valid'

INFO  - Using valid Enterprise Edition CPU license (hivemq-license.lic) issued to HiveMQ - Internal for max 9999 CPU cores, valid until 2024-03-31.
INFO  - Using valid license (kafka-license.elic) for enterprise extension with name "HiveMQ Enterprise Extension for Kafka", valid until 2024-03-31.

2. The HiveMQ log contains info about successful Kafka connection:

kubectl logs deployments/hivemq -f | grep -i kafka

INFO  - Starting extension with id "hivemq-kafka-extension" at /opt/hivemq/extensions/hivemq-kafka-extension
INFO  - No mqtt-to-kafka transformers were found. No mqtt-to-kafka transformers are started.
INFO  - Started mqtt-to-kafka mapping "mapping01" with MQTT topics [#] and Kafka topic "test" for Kafka Cluster "cluster01"
INFO  - No kafka-to-mqtt transformers were found. No kafka kafka-to-mqtt transformers are started.
INFO  - Started kafka-to-mqtt mapping "mapping02" with kafka topic: [test] and kafka pattern: [test-(.)*] for Kafka Cluster "cluster01" (poll duration: 100 ms).
INFO  - Extension "HiveMQ Enterprise Extension for Kafka" version 4.24.0 started successfully.
INFO  - Subscribing kafka-to-mqtt topic-mapping with id mapping02 to kafka pattern (test-(.)*|\Qtest\E)

3. Kafka dashboard is visible in the HiveMQ Control Center:

Testing Message Flow between HiveMQ Broker and Kafka using MQTT CLI

1. Subscribe a reference MQTT client to the topic “test”, The output shows the topic name and message: (please update your hostname with the DNS name or public IP address of your hivemq-hivemq-mqtt service, 20.113.46.120 in our test)

   mqtt subscribe --topic '#' --host 20.113.46.120 --port 1883 -q 1 --showTopics

   Do not close this terminal session!

2. From a different terminal session, publish a message to the topic “test”:(please update your hostname with the DNS name or public IP address of your hivemq-hivemq-mqtt service, 20.113.46.120 in our test)

   mqtt publish --topic test --message Hello --host 20.113.46.120 --port 1883 -q 1

3. If everything is correct, the subscriber will indefinitely receive the message we published. You can stop this by terminating the execution of the command by pressing Ctrl+C.

   mqtt subscribe --topic '#' --host $hivemqhost --port 1883 -q 1 --showTopics
   
   test: Hello
   test: Hello
   test: Hello
   test: Hello
   ........

4. the Kafka Dashboard in the HiveMQ Control Center shows incoming and outgoing Kafka messaging: