Cluster Stack Migration

To be able to start migrating the Cluster Stack, you need to have migrated all Self-Service instances.

This Cluster Stack Migration needs to be executed for each Kafka cluster defined in your Axual installation.

Start a standalone Cluster-API

Objective

Come up with a new values.yaml file to start a new Cluster-API with axual-helm-charts alongside existing Kafka.

Execution

To avoid downtime or customer impact, perform this execution outside office hours

Standalone Cluster-API

Define a new Chart.yaml that defines as dependency the platform:0.17.11 charts.

Chart.yaml

apiVersion: v2
appVersion: "2024.1.3"
description: Cluster API with Axual Helm Charts
name: just-cluster-api
type: application
version: 0.1.0
dependencies:
  - name: "platform"
    version: "0.17.11"
    repository: "https://dev.axual.io/nexus/repository/axual-helm-stable"

Create a new values.yaml using the existing Cluster’s values.yaml used by the Axual Helm Charts as reference.

First, disable the components that are not Cluster-API

values.yaml

global:
  cluster:
    enabled: true
    name: [existing-cluster-name]

    strimzi:
      enabled: false
    clusterbrowse:
      enabled: false

  instance:
    enabled: false

  mgmt:
    enabled: false

In the core.clusterapi section, provide the same configuration used by the existing Cluster’s values.yaml
values.yaml
platform: core: clusterapi: [existing-configuration]

You will not be able to deploy it because the resources will still exist from the existing Cluster stack deployment.

Existing Cluster-API

In the existing Cluster stack deployment, disable the Cluster-API
values.yaml
global: cluster: clusterapi: enabled: false
Upgrade the existing Cluster stack deployment to disable the Cluster-API
Start the new Standalone Cluster deployment to replace the disabled Cluster-API

Perform these two above steps in a fast sequence to reduce to the minimum the amount of time when a restart of Discovery-API and Schema-Registry might fail due to missing Cluster-API.

The Self-Service functionalities are not affected since all the clusters have been made not requiring Cluster-API

Verification

In this step, we are going to verify that the new Cluster-API can access the existing Kafka.

You can verify this be either

Restart a Discovery-API deployment
Restart a Schema-Registry deployment

If all checks are successful, you can proceed to the next steps.

Migrate Axual Operator to the Strimzi Operator

Objective

Change the Operator available in the k8s-cluster which deploys Kafka and Zookeeper pods

Be sure that you are using the same version of the Axual Operator

Execution

Based on your way to deploy the Axual Operator, you might have two different ways.

Helm Upgrade Command

Download the Strimzi public charts

helm repo add strimzi https://strimzi.io/charts/

Upgrade the existing Axual Operator with the new values

helm upgrade --install [existing-release-name] strimzi/strimzi-kafka-operator \
        --version=0.34.0 \
        --namespace [existing-namespace] \
        --set watchAnyNamespace=true \
        --set kafka.image.registry=registry.axual.io \
        --set kafka.image.repository=axual/streaming/strimzi \
        --set image.imagePullSecrets=[existing-docker-secret-name]

Chart.yaml and Values.yaml

Update the existing Chart.yaml to use strimzi-kafka-operator as dependency chart.

Chart.yaml

apiVersion: v2
name: "strimzi-operator"
type: "application"
version: "0.34.0"
appVersion: "2024.1.3"
description: Strimzi Operator
dependencies:
  - name: "strimzi-kafka-operator"
    version: "0.34.0"
    repository: "https://strimzi.io/charts/"

Update the existing values.yaml to pull a different Kafka image from Axual Registry.

values.yaml

strimzi-kafka-operator:
  watchAnyNamespace: true
  createGlobalResources: true
  # Adjust to your needs
  resources:
    limits:
      memory: 512Mi
    requests:
      memory: 512Mi
      cpu: 200m
  image:
    imagePullSecrets: [existing-docker-secret-name]
  kafka:
    image:
      registry: "registry.axual.io"
      repository: "axual/streaming/strimzi"

This will cause rolling restarts of Zookeeper and Kafka pods.

Verification

In this step, we are going to verify that the new Zookeeper and Kafka pods deployed with Strimzi Operator work as the old Zookeeper and Kakfa pods deployed with Axual Operator.

You can verify this be either

Confirming that your producer/consumer applications are running fine
Logging into the Self-Service and perform a topic deployment
Logging into the Self-Service and browse a topic deployment

If all checks are successful, you can proceed to the next steps.

Migrate Kafka

Objective

Make the existing Kafka installation to be deployed using the axual-streaming-charts.

Execution

Before upgrading the Kafka deployment, check the diffs with ArgoCD, with helm diff upgrade --install, or in the way supported from the tool you are using to deploy the charts.

Replace the dependency in the Cluster Chart.yaml file.

Chart.yaml

apiVersion: v2
appVersion: "2024.1.3"
description: Kafka Stack with Streaming Charts
name: [existing-cluster-name]
type: application
version: 0.1.0
dependencies:
  - name: "axual-streaming"
    version: "0.3.4"
    repository: "https://dev.axual.io/nexus/repository/axual-helm-stable"

Copy the existing platform.core.kafka section from the existing values.yaml file so that you can replace some keys.

Disable the components expect kafka

values.yaml

global:

  rest-proxy:
    enabled: false

  apicurio:
    enabled: false

  axual-schema-registry:
    enabled: false

Add the Cluster name to the axual-streaming.kafka.fullnameOverride key
values.yaml
axual-streaming: kafka: fullnameOverride: [existing-cluster-name]
Be sure that to use the right cluster-name to avoid any Kafka restart

Add the Internal Listeners Configuration to the axual-streaming.kafka.kafka key

values.yaml

axual-streaming:
  kafka:
    kafka:
      # Kafka internal listener configuration
     internalListenerTlsEnabled: "true"
     internalListenerAuthenticationType: tls

Replace platform.core.kafka key with axual-streaming.kafka key

existing_values.yaml

platform:
  core:
    kafka:
      [existing-content]

new_values.yaml

axual-streaming:
  kafka:
    [existing-content]

Replace global.cluster.kafka.nodes key with axual-streaming.kafka.kafka.replicas key

existing_values.yaml

global:
  cluster:
    kafka:
      nodes: [existing-value]

new_values.yaml

axual-streaming:
  kafka:
    kafka:
      replicas: [existing-value]

Replace global.cluster.zookeeper.nodes key with axual-streaming.kafka.zookeeper.replicas key

existing_values.yaml

global:
  cluster:
    zookeeper:
      nodes: [existing-value]

new_values.yaml

axual-streaming:
  kafka:
    zookeeper:
      replicas: [existing-value]

Replace platform.core.kafka.rackEnabled key with axual-streaming.kafka.kafka.rack.enabled key

existing_values.yaml

platform:
  core:
    kafka:
      kafka:
        rackEnabled: [existing-value]

new_values.yaml

axual-streaming:
  kafka:
    kafka:
      rack:
        enabled: [existing-value]

Replace platform.core.kafka.rackTopologyKey key with axual-streaming.kafka.kafka.rack.topologyKey key

existing_values.yaml

platform:
  core:
    kafka:
      kafka:
        rackTopologyKey: [existing-value]

new_values.yaml

axual-streaming:
  kafka:
    kafka:
      rack:
        topologyKey: [existing-value]

Replace platform.core.kafka.rackEnabled key with axual-streaming.kafka.zookeeper.rack.enabled key

existing_values.yaml

platform:
  core:
    kafka:
      kafka:
        rackEnabled: [existing-value]

new_values.yaml

axual-streaming:
  kafka:
    zookeeper:
      rack:
        enabled: [existing-value]

Replace platform.core.kafka.rackTopologyKey key with axual-streaming.kafka.zookeeper.rack.topologyKey key

existing_values.yaml

platform:
  core:
    kafka:
      kafka:
        rackTopologyKey: [existing-value]

new_values.yaml

axual-streaming:
  kafka:
    zookeeper:
      rack:
        topologyKey: [existing-value]

Replace platform.core.kafka.superUsers key with axual-streaming.kafka.kafka.authorization.superUsers key

existing_values.yaml

platform:
  core:
    kafka:
      kafka:
        superUsers:
          [existing-value]

new_values.yaml

axual-streaming:
  kafka:
    kafka:
      authorization:
        superUsers:
          [existing-value]

In the Kafka security section, provide the correct values for clientsCaCertGeneration, clientsCaGeneration, clusterCaCertGeneration, clusterCaGeneration

values.yaml

axual-streaming:
    kafka:
        kafka:
            security:
                clientsCaCertGeneration: "0"
                clientsCaGeneration: "0"
                clusterCaCertGeneration: "0"
                clusterCaGeneration: "0"

If the configuration fully matches, there will be no restart of the cluster. If there are differences, keep adapting the values.yaml until all the diffs are gone.

Verification

In this step, we are going to verify that the existing Zookeeper and Kafka pods deployed with Strimzi Operator and Streaming Charts work as expected.

You can verify this be either

Confirming that your producer/consumer applications are running fine
Logging into the Self-Service and perform a topic deployment
Logging into the Self-Service and browse a topic deployment

If all checks are successful, you can proceed to the next steps.