You must have a Kubernetes cluster that has Helm configured.

Tested Software

Install Helm on Kubernetes

helm init
helm repo update
helm list

kubectl delete --namespace kube-system svc tiller-deploy
kubectl delete --namespace kube-system deploy tiller-deploy
kubectl create serviceaccount --namespace kube-system tiller
kubectl create clusterrolebinding tiller-cluster-rule --clusterrole=cluster-admin --serviceaccount=kube-system:tiller
kubectl patch deploy --namespace kube-system tiller-deploy -p '{"spec":{"template":{"spec":{"serviceAccount":"tiller"}}}}'
helm init --service-account tiller --upgrade

The ZooKeeper and Kafka cluster deployed with StatefulSets that have a volumeClaimTemplate which provides the persistent volume for each replica. You can define the size of the volumes by changing dataDirSize and dataLogDirSize under cp-zookeeper and size under cp-kafka in values.yaml.

You also could use the cloud provider’s volumes by specifying StorageClass. For example, if you are on AWS your storage class will look like this:

The StorageClass that was created can be specified in dataLogDirStorageClass and dataDirStorageClass under cp-zookeeper and in storageClass+ under cp-kafka in values.yaml.

To deploy non-persistent Kafka and ZooKeeper clusters, you must change the value of persistence.enabled under cp-kafka and cp-zookeeper in values.yaml

Clone the Confluent Helm Chart repo

Install a 3 node Zookeeper ensemble, a Kafka cluster of 3 brokers, 1 Confluent Schema Registry instance, 1 REST Proxy instance, and 1 Kafka Connect worker, 1 ksqlDB server in your Kubernetes environment.

If you want to install without the Confluent Schema Registry instance, the REST Proxy instance, and the Kafka Connect worker:

View the installed Helm releases:

Using Helm

helm test my-confluent-oss

Verify Kafka cluster

kubectl get pods

kubectl exec -c cp-kafka-broker -it my-confluent-oss-cp-kafka-0 -- /bin/bash /usr/bin/kafka-console-producer --broker-list localhost:9092 --topic test

m1
m2

kubectl exec -c cp-kafka-broker -it my-confluent-oss-cp-kafka-0 -- /bin/bash  /usr/bin/kafka-console-consumer --bootstrap-server localhost:9092 --topic test --from-beginning

Manual Test

git clone https://github.com/confluentinc/cp-helm-charts.git        #(1)
kubectl apply -f cp-helm-charts/examples/zookeeper-client.yaml      #(2)
...
kubectl exec -it zookeeper-client -- /bin/bash zookeeper-shell <zookeeper service>:<port> ls /brokers/ids       #(3)
kubectl exec -it zookeeper-client -- /bin/bash zookeeper-shell <zookeeper service>:<port> get /brokers/ids/0
kubectl exec -it zookeeper-client -- /bin/bash zookeeper-shell <zookeeper service>:<port> ls /brokers/topics    #(4)

kubectl apply -f cp-helm-charts/examples/kafka-client.yaml #(1)
kubectl exec -it kafka-client -- /bin/bash      #(2)

## Setup
export RELEASE_NAME=<release name>
export ZOOKEEPERS=${RELEASE_NAME}-cp-zookeeper:2181
export KAFKAS=${RELEASE_NAME}-cp-kafka-headless:9092

## Create Topic
kafka-topics --zookeeper $ZOOKEEPERS --create --topic test-rep-one --partitions 6 --replication-factor 1

## Producer
kafka-run-class org.apache.kafka.tools.ProducerPerformance --print-metrics --topic test-rep-one --num-records 6000000 --throughput 100000 --record-size 100 --producer-props bootstrap.servers=$KAFKAS buffer.memory=67108864 batch.size=8196

## Consumer
kafka-consumer-perf-test --broker-list $KAFKAS --messages 6000000 --threads 1 --topic test-rep-one --print-metrics

Run A Streams Application

Scaling Zookeeper

helm install cp-helm-charts

helm upgrade <release name> cp-helm-charts

helm upgrade <release name> cp-helm-charts

Scaling Kafka

helm install cp-helm-charts

helm upgrade <release name> cp-helm-charts

helm upgrade <release name> cp-helm-charts

Monitoring

To remove the pods, list the pods with kubectl get pods and then delete the pods by name.

To delete the Helm release, find the Helm release name with helm list and delete it with helm delete. You may also need to clean up leftover StatefulSets, since helm delete can leave them behind. Finally, clean up all persisted volume claims (pvc) created by this release.

Minikube version 0.23.0 or higher is required for docker server moby/moby#31352], which adds support for using ARG in FROM in your Dockerfile.

First follow the basic Minikube installation instructions.

Then install the Minikube drivers. Minikube uses Docker Machine to manage the Kubernetes VM so it benefits from the driver plugin architecture that Docker Machine uses to provide a consistent way to manage various VM providers. Minikube embeds VirtualBox and VMware Fusion drivers so there are no additional steps to use them. However, other drivers require an extra binary to be present in the host PATH.

1. Start Minikube. The following command increases the memory to 6096 MB and uses the xhyve driver for the native macOS Hypervisor.

   minikube start --kubernetes-version v1.9.4 --cpus 4 --memory 6096 --vm-driver=xhyve --v=8

2. Continue to check status of your local Kubernetes cluster until both minikube and cluster are in Running state

   ❯ minikube status
   m01
   host: Running
   kubelet: Running
   apiserver: Running
   kubeconfig: Configured

3. Work around Minikube issue #1568.

   minikube ssh -- sudo ip link set docker0 promisc on

4. Set the context.

   eval $(minikube docker-env)
   
   kubectl config set-context minikube.internal --cluster=minikube --user=minikube
   Context "minikube.internal" modified.
   
   kubectl config use-context minikube.internal
   Switched to context "minikube.internal".

Verify Minikube Local Kubernetes Environment

kubectl config current-context
minikube.internal

kubectl cluster-info
Kubernetes master is running at https://192.168.99.106:8443
KubeDNS is running at https://192.168.99.106:8443/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy