DEV Community

Gilson Oliveira
Gilson Oliveira

Posted on

Don't Let Bitnami Changes Break Your Kafka Pipelines: Migrate to KRaft Mode Now

#kafka #devops #docker #programming

Bitnami is deprecating their Kafka images, and ZooKeeper is being removed from Apache Kafka 4.0. Time to migrate to KRaft mode with Confluent's official images for a future-proof, simpler architecture.

If you're using bitnami/kafka in your Docker setup, you've probably seen this warning:

⚠️ Important Notice: Beginning August 28th, 2025, Bitnami will evolve its public catalog... All existing container images have been migrated to "Bitnami Legacy" repository where they will no longer receive updates.

But there's a bigger issue: Apache Kafka is removing ZooKeeper entirely in version 4.0. Most current setups still rely on ZooKeeper, which means double trouble ahead.

The Solution: KRaft Mode

KRaft (Kafka Raft) eliminates ZooKeeper dependency entirely, giving you:

✅ Simpler architecture (fewer moving parts)
✅ Better performance and scalability
✅ Future-proof (ZooKeeper support ends with Kafka 3.9)
✅ Faster startup and recovery times

Before: The Old Way

version: '3.8'
services:
  zookeeper:
    image: bitnami/zookeeper:latest
    ports:
      - "2181:2181"
    environment:
      - ALLOW_ANONYMOUS_LOGIN=yes

  kafka:
    image: bitnami/kafka:latest
    ports:
      - "9092:9092"
    environment:
      - KAFKA_BROKER_ID=1
      - KAFKA_CFG_ZOOKEEPER_CONNECT=zookeeper:2181
      - ALLOW_PLAINTEXT_LISTENER=yes
      - KAFKA_CFG_LISTENER_SECURITY_PROTOCOL_MAP=PLAINTEXT:PLAINTEXT
      - KAFKA_CFG_LISTENERS=PLAINTEXT://:9092
      - KAFKA_CFG_ADVERTISED_LISTENERS=PLAINTEXT://localhost:9092
    depends_on:
      - zookeeper
Enter fullscreen mode Exit fullscreen mode

Problems:

  • Deprecated Bitnami images (no more updates)
  • ZooKeeper dependency (removed in Kafka 4.0)
  • Two services to manage

After: The Modern KRaft Way
Here's the future-proof solution using Confluent's official image:

version: '3.8'
services:
  kafka:
    image: confluentinc/cp-kafka:7.4.0
    ports:
      - "9092:9092"
    environment:
      # KRaft Configuration
      KAFKA_NODE_ID: 1
      KAFKA_PROCESS_ROLES: broker,controller
      KAFKA_CONTROLLER_QUORUM_VOTERS: 1@kafka:9093
      KAFKA_CONTROLLER_LISTENER_NAMES: CONTROLLER
      CLUSTER_ID: MkU3OEVBNTcwNTJENDM1Tk

      # Network Configuration  
      KAFKA_LISTENER_SECURITY_PROTOCOL_MAP: CONTROLLER:PLAINTEXT,PLAINTEXT:PLAINTEXT
      KAFKA_LISTENERS: PLAINTEXT://0.0.0.0:9092,CONTROLLER://0.0.0.0:9093
      KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://kafka:9092

      # Storage Configuration
      KAFKA_LOG_DIRS: /tmp/kraft-combined-logs

      # Topic Configuration
      KAFKA_OFFSETS_TOPIC_REPLICATION_FACTOR: 1
      KAFKA_TRANSACTION_STATE_LOG_REPLICATION_FACTOR: 1
      KAFKA_TRANSACTION_STATE_LOG_MIN_ISR: 1
      KAFKA_AUTO_CREATE_TOPICS_ENABLE: "true"
Enter fullscreen mode Exit fullscreen mode

Key Configuration Explained

Let me break down the important KRaft settings:

KAFKA_NODE_ID: 1                           # Unique node identifier
KAFKA_PROCESS_ROLES: broker,controller     # Combined mode (handles both roles)
KAFKA_CONTROLLER_QUORUM_VOTERS: 1@kafka:9093  # Controller election setup
CLUSTER_ID: MkU3OEVBNTcwNTJENDM2Qk         # Unique cluster identifier
Enter fullscreen mode Exit fullscreen mode

Network Configuration

KAFKA_LISTENERS: PLAINTEXT://0.0.0.0:9092,CONTROLLER://0.0.0.0:9093
KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://kafka:9092
KAFKA_CONTROLLER_LISTENER_NAMES: CONTROLLER
Enter fullscreen mode Exit fullscreen mode

The controller listener (port 9093) is used internally for KRaft consensus, while 9092 remains your application port.

Benefits You'll See Immediately

  1. Faster Startup: No waiting for ZooKeeper election
  2. Simpler Debugging: One service instead of two
  3. Better Resource Usage: No ZooKeeper overhead
  4. Future-Proof: Ready for Kafka 4.0

Conclusion

The writing is on the wall: ZooKeeper is going away, and Bitnami is changing their strategy. Don't wait for your pipelines to break in production.
KRaft mode isn't just a workaround, it's genuinely better. Simpler architecture, better performance. Your future self will thank you for making this change now.

Have you made the switch to KRaft? let me know your experience in the comments!

Top comments (2)

Collapse
 
ivis1 profile image
Ivan Isaac •

Great write-up! One tip: generate/format KRaft storage before first start (kafka-storage.sh random-uuid; kafka-storage.sh format --cluster-id … --config …) and mount a volume for KAFKA_LOG_DIRS so the cluster ID survives restarts. Also pin cp-kafka to a specific version and, if you scale to 3 brokers, list all voters in KAFKA_CONTROLLER_QUORUM_VOTERS and bump replication factors. Did you run into any advertised.listeners quirks with Docker host vs service name?

Collapse
 
gilson_oliveira profile image
Gilson Oliveira •

Hey Ivan, thanks for the tips! You’re right about formatting storage and persisting KAFKA_LOG_DIRS, that’s a great call for production. I’ve only tested using the service name in advertised.listeners, didn’t face issues so far, did you run into problems with host vs service name?