kafka-kubernetes（Kafka Kubernetes 部署）

Kafka on Kubernetes Deployment

Expert guidance for deploying Apache Kafka on Kubernetes using industry-standard tools.

When to Use This Skill

I activate when you need help with:

• Kubernetes deployments: "Deploy Kafka on Kubernetes", "run Kafka in K8s", "Kafka Helm chart"
• Operator selection: "Strimzi vs Confluent Operator", "which Kafka operator to use"
• StatefulSet patterns: "Kafka StatefulSet best practices", "persistent volumes for Kafka"
• Production K8s: "Production-ready Kafka on K8s", "Kafka high availability in Kubernetes"

What I Know

Deployment Options Comparison

Recommendation: Strimzi Operator for most production use cases (CNCF project, active community, KRaft support)

Deployment Approach 1: Strimzi Operator (Recommended)

Strimzi is a CNCF Sandbox project providing Kubernetes operators for Apache Kafka.

Features

• ✅ KRaft mode support (Kafka 3.6+, no ZooKeeper)
• ✅ Declarative Kafka management (CRDs)
• ✅ Automatic rolling upgrades
• ✅ Built-in monitoring (Prometheus metrics)
• ✅ Mirror Maker 2 for replication
• ✅ Kafka Connect integration
• ✅ User and topic management via CRDs

Installation (Helm)

# 1. Add Strimzi Helm repository
helm repo add strimzi https://strimzi.io/charts/
helm repo update

# 2. Create namespace
kubectl create namespace kafka

# 3. Install Strimzi Operator
helm install strimzi-kafka-operator strimzi/strimzi-kafka-operator \
  --namespace kafka \
  --set watchNamespaces="{kafka}" \
  --version 0.39.0

# 4. Verify operator is running
kubectl get pods -n kafka
# Output: strimzi-cluster-operator-... Running

Deploy Kafka Cluster (KRaft Mode)

# kafka-cluster.yaml
apiVersion: kafka.strimzi.io/v1beta2
kind: KafkaNodePool
metadata:
  name: kafka-pool
  namespace: kafka
  labels:
    strimzi.io/cluster: my-kafka-cluster
spec:
  replicas: 3
  roles:
    - controller
    - broker
  storage:
    type: jbod
    volumes:
      - id: 0
        type: persistent-claim
        size: 100Gi
        class: fast-ssd
        deleteClaim: false
---
apiVersion: kafka.strimzi.io/v1beta2
kind: Kafka
metadata:
  name: my-kafka-cluster
  namespace: kafka
  annotations:
    strimzi.io/kraft: enabled
    strimzi.io/node-pools: enabled
spec:
  kafka:
    version: 3.7.0
    metadataVersion: 3.7-IV4
    replicas: 3

    listeners:
      - name: plain
        port: 9092
        type: internal
        tls: false
      - name: tls
        port: 9093
        type: internal
        tls: true
        authentication:
          type: tls
      - name: external
        port: 9094
        type: loadbalancer
        tls: true
        authentication:
          type: tls

    config:
      default.replication.factor: 3
      min.insync.replicas: 2
      offsets.topic.replication.factor: 3
      transaction.state.log.replication.factor: 3
      transaction.state.log.min.isr: 2
      auto.create.topics.enable: false
      log.retention.hours: 168
      log.segment.bytes: 1073741824
      compression.type: lz4

    resources:
      requests:
        memory: 4Gi
        cpu: "2"
      limits:
        memory: 8Gi
        cpu: "4"

    jvmOptions:
      -Xms: 2048m
      -Xmx: 4096m

    metricsConfig:
      type: jmxPrometheusExporter
      valueFrom:
        configMapKeyRef:
          name: kafka-metrics
          key: kafka-metrics-config.yml

# Apply Kafka cluster
kubectl apply -f kafka-cluster.yaml

# Wait for cluster to be ready (5-10 minutes)
kubectl wait kafka/my-kafka-cluster --for=condition=Ready --timeout=600s -n kafka

# Check status
kubectl get kafka -n kafka
# Output: my-kafka-cluster   3.7.0   3         True

Create Topics (Declaratively)

# kafka-topics.yaml
apiVersion: kafka.strimzi.io/v1beta2
kind: KafkaTopic
metadata:
  name: user-events
  namespace: kafka
  labels:
    strimzi.io/cluster: my-kafka-cluster
spec:
  partitions: 12
  replicas: 3
  config:
    retention.ms: 604800000  # 7 days
    segment.bytes: 1073741824
    compression.type: lz4
---
apiVersion: kafka.strimzi.io/v1beta2
kind: KafkaTopic
metadata:
  name: order-events
  namespace: kafka
  labels:
    strimzi.io/cluster: my-kafka-cluster
spec:
  partitions: 6
  replicas: 3
  config:
    retention.ms: 2592000000  # 30 days
    min.insync.replicas: 2

# Apply topics
kubectl apply -f kafka-topics.yaml

# Verify topics created
kubectl get kafkatopics -n kafka

Create Users (Declaratively)

# kafka-users.yaml
apiVersion: kafka.strimzi.io/v1beta2
kind: KafkaUser
metadata:
  name: my-producer
  namespace: kafka
  labels:
    strimzi.io/cluster: my-kafka-cluster
spec:
  authentication:
    type: tls
  authorization:
    type: simple
    acls:
      - resource:
          type: topic
          name: user-events
          patternType: literal
        operations: [Write, Describe]
      - resource:
          type: topic
          name: order-events
          patternType: literal
        operations: [Write, Describe]
---
apiVersion: kafka.strimzi.io/v1beta2
kind: KafkaUser
metadata:
  name: my-consumer
  namespace: kafka
  labels:
    strimzi.io/cluster: my-kafka-cluster
spec:
  authentication:
    type: tls
  authorization:
    type: simple
    acls:
      - resource:
          type: topic
          name: user-events
          patternType: literal
        operations: [Read, Describe]
      - resource:
          type: group
          name: my-consumer-group
          patternType: literal
        operations: [Read]

# Apply users
kubectl apply -f kafka-users.yaml

# Get user credentials (TLS certificates)
kubectl get secret my-producer -n kafka -o jsonpath='{.data.user\.crt}' | base64 -d > producer.crt
kubectl get secret my-producer -n kafka -o jsonpath='{.data.user\.key}' | base64 -d > producer.key
kubectl get secret my-kafka-cluster-cluster-ca-cert -n kafka -o jsonpath='{.data.ca\.crt}' | base64 -d > ca.crt

Deployment Approach 2: Confluent Operator

Confluent for Kubernetes (CFK) provides enterprise-grade Kafka management.

Features

• ✅ Full Confluent Platform (Kafka, Schema Registry, ksqlDB, Connect)
• ✅ Hybrid deployments (K8s + on-prem)
• ✅ Rolling upgrades with zero downtime
• ✅ Multi-region replication
• ✅ Advanced security (RBAC, encryption)
• ⚠️ Requires Confluent Platform license (paid)

Installation

# 1. Add Confluent Helm repository
helm repo add confluentinc https://packages.confluent.io/helm
helm repo update

# 2. Create namespace
kubectl create namespace confluent

# 3. Install Confluent Operator
helm install confluent-operator confluentinc/confluent-for-kubernetes \
  --namespace confluent \
  --version 0.921.11

# 4. Verify
kubectl get pods -n confluent

Deploy Kafka Cluster

# kafka-cluster-confluent.yaml
apiVersion: platform.confluent.io/v1beta1
kind: Kafka
metadata:
  name: kafka
  namespace: confluent
spec:
  replicas: 3
  image:
    application: confluentinc/cp-server:7.6.0
    init: confluentinc/confluent-init-container:2.7.0

  dataVolumeCapacity: 100Gi
  storageClass:
    name: fast-ssd

  metricReporter:
    enabled: true

  listeners:
    internal:
      authentication:
        type: plain
      tls:
        enabled: true
    external:
      authentication:
        type: plain
      tls:
        enabled: true

  dependencies:
    zookeeper:
      endpoint: zookeeper.confluent.svc.cluster.local:2181

  podTemplate:
    resources:
      requests:
        memory: 4Gi
        cpu: 2
      limits:
        memory: 8Gi
        cpu: 4

# Apply Kafka cluster
kubectl apply -f kafka-cluster-confluent.yaml

# Wait for cluster
kubectl wait kafka/kafka --for=condition=Ready --timeout=600s -n confluent

Deployment Approach 3: Bitnami Helm Chart (Dev/Staging)

Bitnami Helm Chart is simple but less suitable for production.

Installation

# 1. Add Bitnami repository
helm repo add bitnami https://charts.bitnami.com/bitnami
helm repo update

# 2. Install Kafka (KRaft mode)
helm install kafka bitnami/kafka \
  --namespace kafka \
  --create-namespace \
  --set kraft.enabled=true \
  --set controller.replicaCount=3 \
  --set broker.replicaCount=3 \
  --set persistence.size=100Gi \
  --set persistence.storageClass=fast-ssd \
  --set metrics.kafka.enabled=true \
  --set metrics.jmx.enabled=true

# 3. Get bootstrap servers
export KAFKA_BOOTSTRAP=$(kubectl get svc kafka -n kafka -o jsonpath='{.status.loadBalancer.ingress[0].hostname}'):9092

Limitations:

• ⚠️ Less production-ready than Strimzi/Confluent
• ⚠️ Limited declarative topic/user management
• ⚠️ Fewer advanced features (no MirrorMaker 2, limited RBAC)

Production Best Practices

1. Storage Configuration

Use SSD-backed storage classes for Kafka logs:

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: fast-ssd
provisioner: kubernetes.io/aws-ebs  # or pd.csi.storage.gke.io for GKE
parameters:
  type: gp3  # AWS EBS GP3 (or io2 for extreme performance)
  iopsPerGB: "50"
  throughput: "125"
allowVolumeExpansion: true
volumeBindingMode: WaitForFirstConsumer

Kafka storage requirements:

• Min IOPS: 3000+ per broker
• Min Throughput: 125 MB/s per broker
• Persistent: Use deleteClaim: false (don't delete data on pod deletion)

2. Resource Limits

resources:
  requests:
    memory: 4Gi
    cpu: "2"
  limits:
    memory: 8Gi
    cpu: "4"

jvmOptions:
  -Xms: 2048m  # Initial heap (50% of memory request)
  -Xmx: 4096m  # Max heap (50% of memory limit, leave room for OS cache)

Sizing guidelines:

• Small (dev): 2 CPU, 4Gi memory
• Medium (staging): 4 CPU, 8Gi memory
• Large (production): 8 CPU, 16Gi memory

3. Pod Disruption Budgets

Ensure high availability during K8s upgrades:

apiVersion: policy/v1
kind: PodDisruptionBudget
metadata:
  name: kafka-pdb
  namespace: kafka
spec:
  maxUnavailable: 1
  selector:
    matchLabels:
      app.kubernetes.io/name: kafka

4. Affinity Rules

Spread brokers across availability zones:

spec:
  kafka:
    template:
      pod:
        affinity:
          podAntiAffinity:
            requiredDuringSchedulingIgnoredDuringExecution:
              - labelSelector:
                  matchExpressions:
                    - key: strimzi.io/name
                      operator: In
                      values:
                        - my-kafka-cluster-kafka
                topologyKey: topology.kubernetes.io/zone

5. Network Policies

Restrict access to Kafka brokers:

apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: kafka-network-policy
  namespace: kafka
spec:
  podSelector:
    matchLabels:
      strimzi.io/name: my-kafka-cluster-kafka
  policyTypes:
    - Ingress
  ingress:
    - from:
      - podSelector:
          matchLabels:
            app: my-producer
      - podSelector:
          matchLabels:
            app: my-consumer
      ports:
      - protocol: TCP
        port: 9092
      - protocol: TCP
        port: 9093

Monitoring Integration

Prometheus + Grafana Setup

Strimzi provides built-in Prometheus metrics exporter:

# kafka-metrics-configmap.yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: kafka-metrics
  namespace: kafka
data:
  kafka-metrics-config.yml: |
    # Use JMX Exporter config from:
    # plugins/specweave-kafka/monitoring/prometheus/kafka-jmx-exporter.yml
    lowercaseOutputName: true
    lowercaseOutputLabelNames: true
    whitelistObjectNames:
      - "kafka.server:type=BrokerTopicMetrics,name=*"
      # ... (copy from kafka-jmx-exporter.yml)

# Apply metrics config
kubectl apply -f kafka-metrics-configmap.yaml

# Install Prometheus Operator (if not already installed)
helm install prometheus prometheus-community/kube-prometheus-stack \
  --namespace monitoring \
  --create-namespace

# Create PodMonitor for Kafka
kubectl apply -f - <<EOF
apiVersion: monitoring.coreos.com/v1
kind: PodMonitor
metadata:
  name: kafka-metrics
  namespace: kafka
spec:
  selector:
    matchLabels:
      strimzi.io/kind: Kafka
  podMetricsEndpoints:
    - port: tcp-prometheus
      interval: 30s
EOF

# Access Grafana dashboards (from kafka-observability skill)
kubectl port-forward -n monitoring svc/prometheus-grafana 3000:80
# Open: http://localhost:3000
# Dashboards: Kafka Cluster Overview, Broker Metrics, Consumer Lag, Topic Metrics, JVM Metrics

Troubleshooting

"Pods stuck in Pending state"

Cause: Insufficient resources or storage class not found Fix:

# Check events
kubectl describe pod kafka-my-kafka-cluster-0 -n kafka

# Check storage class exists
kubectl get storageclass

# If missing, create fast-ssd storage class (see Production Best Practices above)

"Kafka broker not ready after 10 minutes"

Cause: Slow storage provisioning or resource limits too low Fix:

# Check broker logs
kubectl logs kafka-my-kafka-cluster-0 -n kafka

# Common issues:
# 1. Low IOPS on storage → Use GP3 or better
# 2. Low memory → Increase resources.requests.memory
# 3. KRaft quorum not formed → Check all brokers are running

"Cannot connect to Kafka from outside K8s"

Cause: External listener not configured Fix:

# Add external listener (Strimzi)
spec:
  kafka:
    listeners:
      - name: external
        port: 9094
        type: loadbalancer
        tls: true
        authentication:
          type: tls

# Get external bootstrap server
kubectl get kafka my-kafka-cluster -n kafka -o jsonpath='{.status.listeners[?(@.name=="external")].bootstrapServers}'

Scaling Operations

Horizontal Scaling (Add Brokers)

# Strimzi: Update KafkaNodePool replicas
kubectl patch kafkanodepool kafka-pool -n kafka --type='json' \
  -p='[{"op": "replace", "path": "/spec/replicas", "value": 5}]'

# Confluent: Update Kafka CR
kubectl patch kafka kafka -n confluent --type='json' \
  -p='[{"op": "replace", "path": "/spec/replicas", "value": 5}]'

# Wait for new brokers
kubectl rollout status statefulset/kafka-my-kafka-cluster-kafka -n kafka

Vertical Scaling (Change Resources)

# Update resources in Kafka CR
kubectl patch kafka my-kafka-cluster -n kafka --type='json' \
  -p='[
    {"op": "replace", "path": "/spec/kafka/resources/requests/memory", "value": "8Gi"},
    {"op": "replace", "path": "/spec/kafka/resources/requests/cpu", "value": "4"}
  ]'

# Rolling restart will happen automatically

Integration with Other Skills

• kafka-iac-deployment: Alternative to K8s (use Terraform for cloud-managed Kafka)
• kafka-observability: Set up Prometheus + Grafana dashboards for K8s Kafka
• kafka-architecture: Cluster sizing and partitioning strategy
• kafka-cli-tools: Test K8s Kafka cluster with kcat

Quick Reference Commands

# Strimzi
kubectl get kafka -n kafka                    # List Kafka clusters
kubectl get kafkatopics -n kafka              # List topics
kubectl get kafkausers -n kafka               # List users
kubectl logs kafka-my-kafka-cluster-0 -n kafka  # Check broker logs

# Confluent
kubectl get kafka -n confluent                # List Kafka clusters
kubectl get schemaregistry -n confluent       # List Schema Registry
kubectl get ksqldb -n confluent               # List ksqlDB

# Port-forward for testing
kubectl port-forward -n kafka svc/my-kafka-cluster-kafka-bootstrap 9092:9092

Next Steps After K8s Deployment:

1. Use kafka-observability skill to verify Prometheus metrics and Grafana dashboards
2. Use kafka-cli-tools skill to test cluster with kcat
3. Deploy your producer/consumer applications to K8s
4. Set up GitOps for declarative topic/user management (ArgoCD, Flux)