How to specify advertised listeners for Kafka multi broker setup on kubernetes and expose the cluster expernally? - apache-kafka

Question

I am trying to setup a multi broker kafka on a kubernetes cluster hosted in Azure. I have a single broker setup working. For the multi broker setup, currently I have an ensemble of zookeeper nodes(3) that manage the kafka service. I am deploying the kafka cluster as a replication controller with replication factor of 3. That is 3 brokers. How can I register the three brokers with Zookeeper such that they register different IP addresses with the Zookeeper?
I bring up my replication controller after the service is deployed and use the Cluster IP in my replication-controller yaml file to specify two advertised.listeners, one for SSL and another for PLAINTEXT. However, in this scenario all brokers register with the same IP and write to replicas fail. I don't want to deploy each broker as a separate replication controller/pod and service as scaling becomes an issue. I would really appreciate any thoughts/ideas on this.
Edit 1:
I am additionally trying to expose the cluster to another VPC in cloud. I have to expose SSL and PLAINTEXT ports for clients which I am doing using advertised.listeners. If I use a statefulset with replication factor of 3 and let kubernetes expose the canonical host names of the pods as host names, these cannot be resolved from an external client. The only way I got this working is to use/expose an external service corresponding to each broker. However, this does not scale.

Answer 1

Kubernetes has the concept of Statefulsets to solve these issues. Each instance of a statefulset has it's own DNS name so you can reference to each instance by a dns name.
This concept is described here in more detail. You can also take a look at this complete example:
apiVersion: v1
kind: Service
metadata:
name: zk-headless
labels:
app: zk-headless
spec:
ports:
- port: 2888
name: server
- port: 3888
name: leader-election
clusterIP: None
selector:
app: zk
---
apiVersion: v1
kind: ConfigMap
metadata:
name: zk-config
data:
ensemble: "zk-0;zk-1;zk-2"
jvm.heap: "2G"
tick: "2000"
init: "10"
sync: "5"
client.cnxns: "60"
snap.retain: "3"
purge.interval: "1"
---
apiVersion: policy/v1beta1
kind: PodDisruptionBudget
metadata:
name: zk-budget
spec:
selector:
matchLabels:
app: zk
minAvailable: 2
---
apiVersion: apps/v1beta1
kind: StatefulSet
metadata:
name: zk
spec:
serviceName: zk-headless
replicas: 3
template:
metadata:
labels:
app: zk
annotations:
pod.alpha.kubernetes.io/initialized: "true"
spec:
affinity:
podAntiAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
- labelSelector:
matchExpressions:
- key: "app"
operator: In
values:
- zk-headless
topologyKey: "kubernetes.io/hostname"
containers:
- name: k8szk
imagePullPolicy: Always
image: gcr.io/google_samples/k8szk:v1
resources:
requests:
memory: "4Gi"
cpu: "1"
ports:
- containerPort: 2181
name: client
- containerPort: 2888
name: server
- containerPort: 3888
name: leader-election
env:
- name : ZK_ENSEMBLE
valueFrom:
configMapKeyRef:
name: zk-config
key: ensemble
- name : ZK_HEAP_SIZE
valueFrom:
configMapKeyRef:
name: zk-config
key: jvm.heap
- name : ZK_TICK_TIME
valueFrom:
configMapKeyRef:
name: zk-config
key: tick
- name : ZK_INIT_LIMIT
valueFrom:
configMapKeyRef:
name: zk-config
key: init
- name : ZK_SYNC_LIMIT
valueFrom:
configMapKeyRef:
name: zk-config
key: tick
- name : ZK_MAX_CLIENT_CNXNS
valueFrom:
configMapKeyRef:
name: zk-config
key: client.cnxns
- name: ZK_SNAP_RETAIN_COUNT
valueFrom:
configMapKeyRef:
name: zk-config
key: snap.retain
- name: ZK_PURGE_INTERVAL
valueFrom:
configMapKeyRef:
name: zk-config
key: purge.interval
- name: ZK_CLIENT_PORT
value: "2181"
- name: ZK_SERVER_PORT
value: "2888"
- name: ZK_ELECTION_PORT
value: "3888"
command:
- sh
- -c
- zkGenConfig.sh && zkServer.sh start-foreground
readinessProbe:
exec:
command:
- "zkOk.sh"
initialDelaySeconds: 15
timeoutSeconds: 5
livenessProbe:
exec:
command:
- "zkOk.sh"
initialDelaySeconds: 15
timeoutSeconds: 5
volumeMounts:
- name: datadir
mountPath: /var/lib/zookeeper
securityContext:
runAsUser: 1000
fsGroup: 1000
volumeClaimTemplates:
- metadata:
name: datadir
spec:
accessModes: [ "ReadWriteOnce" ]
resources:
requests:
storage: 20Gi