Promote Sign in Sign up

Repository

koperator

Oh no! Yet another Kafka operator for Kubernetes

Adobe
Repository

README

Koperator
Released
License
Go version (latest release)

Go version
Go Report Card
CI
CI
Image
Helm chart

Koperator

Koperator is an open-source operator that automates the provisioning, management, and autoscaling of Apache Kafka clusters on Kubernetes.
Unlike other solutions that rely on StatefulSets, Koperator has been built with a unique architecture that provides greater flexibility and functionality for managing Apache Kafka. This architecture allows for fine-grained configuration and management of individual brokers.

Some of the main features of Koperator are:

  • the provisioning of secure and production-ready Kafka clusters
  • fine-grained broker-by-broker configuration support
  • advanced and highly configurable external access
  • graceful Kafka cluster scaling and rebalancing
  • detailed Prometheus metrics
  • encrypted communication using SSL
  • automatic reaction and self-healing based on alerts using Cruise Control
  • graceful rolling upgrades
  • advanced topic and user management via Kubernetes Custom Resources
  • Cruise Control task management via Kubernetes Custom Resources

Architecture

Kafka is a stateful application, and the Kafka Broker is a server that can create and form a cluster with other Brokers. Each Broker has its own unique configuration, the most important of which is the unique broker ID.

Most Kubernetes operators that manage Kafka rely on StatefulSets to create a Kafka Cluster.
While StatefulSets provide unique Broker IDs generated during Pod startup, networking between brokers with headless services, and unique Persistent Volumes for Brokers, they have a few restrictions. For example, Broker configurations cannot be modified independently, and a specific Broker cannot be removed from the cluster - a StatefulSet always removes the most recently created Broker. Furthermore, multiple, different Persistent Volumes cannot be used for each Broker.

Koperator takes a different approach by using simple Pods, ConfigMaps, and PersistentVolumeClaims instead of StatefulSets. These resources allow us to build an Operator that is better suited to manage Apache Kafka.
With Koperator, you can modify the configuration of unique Brokers, remove specific Brokers from clusters, and use multiple Persistent Volumes for each Broker.

If you want to learn more about our design motivations and the scenarios that drove us to create Koperator, please continue reading on our documentation page.

Koperator architecture

Quick start

This quick start guide will walk you through the process of deploying Koperator on an existing Kubernetes cluster and provisioning a Kafka cluster using its custom resources.

Prerequisites

To complete this guide, you will need a Kubernetes cluster (with a suggested minimum of 6 vCPUs and 8 GB RAM). You can run the cluster locally using Kind or Minikube.

The quick start will help you set up a functioning Kafka cluster on Kubernetes. However, it does not include guidance on the installation of Prometheus and cert-manager, which are necessary for some of the more advanced functionality.

Install ZooKeeper

The version of Kafka that is installed by the operator requires Apache ZooKeeper. You'll need to deploy a ZooKeeper cluster if you don’t already have one.

  1. Install ZooKeeper using Pravega’s Zookeeper Operator.
helm install zookeeper-operator --repo https://charts.pravega.io zookeeper-operator --namespace=zookeeper --create-namespace
  1. Create a ZooKeeper cluster.
kubectl create -f -  📦 **View available versions**: [ghcr.io/adobe/koperator/kafka-operator](https://github.com/adobe/koperator/pkgs/container/koperator%2Fkafka-operator/versions)

```sh
# Install the latest release
helm install kafka-operator oci://ghcr.io/adobe/helm-charts/kafka-operator --namespace=kafka --create-namespace

# Or install a specific version (replace with desired version)
helm install kafka-operator oci://ghcr.io/adobe/helm-charts/kafka-operator --version 0.28.0-adobe-20250923 --namespace=kafka --create-namespace

Pull and inspect the chart before installation

# Pull the chart locally
helm pull oci://ghcr.io/adobe/koperator/kafka-operator --version 0.28.0-adobe-20250923

# Extract and inspect
tar -xzf kafka-operator-0.28.0-adobe-20250923.tgz
helm template kafka-operator ./kafka-operator/

# Install from local chart
helm install kafka-operator ./kafka-operator/ --namespace=kafka --create-namespace
  1. Create the Kafka cluster using the KafkaCluster custom resource. The quick start uses a minimal custom resource, but there are other examples in the same directory.
kubectl create -n kafka -f https://raw.githubusercontent.com/adobe/koperator/master/config/samples/simplekafkacluster.yaml
  1. Verify that the Kafka cluster has been created.
> kubectl get pods -n kafka

kafka-0-nvx8c                             1/1     Running   0          16m
kafka-1-swps9                             1/1     Running   0          15m
kafka-2-lppzr                             1/1     Running   0          15m
kafka-cruisecontrol-fb659b84b-7cwpn       1/1     Running   0          15m
kafka-operator-operator-8bb75c7fb-7w4lh   2/2     Running   0          17m

Test Kafka cluster

To test the Kafka cluster let's create a topic and send some messages.

  1. You can use the KafkaTopic CR to create a topic called my-topic:

kubectl create -n kafka -f -