布置 Raft Kafka(Kafka 3.3.1 及以上版别引入的 KRaft 形式)在 Kubernetes (k8s) 上,能够简化 Kafka 集群的管理,由于它不再依赖于 Zookeeper。
关于Raft Kafka 介绍和物理机布置能够参阅我以下文章:
以下是布置 Raft Kafka 集群的基本步骤:
1)预备 Kubernetes 集群
确保你有一个运行中的 Kubernetes 集群,并且已经装备了 kubectl 命令行工具。 布置教程如下:
2)装置 helm
# 下载包
wget https://get.helm.sh/helm-v3.9.4-linux-amd64.tar.gz
# 解压压缩包
tar -xf helm-v3.9.4-linux-amd64.tar.gz
# 制作软衔接
ln -s /opt/helm/linux-amd64/helm /usr/local/bin/helm
# 验证
helm version
helm help
3)装备 Helm chart
如果你运用 Bitnami 的 Kafka Helm chart,你需求创立一个 values.yaml
文件来装备 Kafka 集群。在该文件中,你能够启用 KRaft 形式并装备其他设置,如认证、端口等。
# 增加下载源
helm repo add bitnami https://charts.bitnami.com/bitnami
# 下载
helm pull bitnami/kafka --version 26.0.0
# 解压
tar -xf kafka-26.0.0.tgz
# 修正装备
vi kafka/values.yaml
以下是一个 values.yaml 的示例装备:
image:
registry: registry.cn-hangzhou.aliyuncs.com
repository: bigdata_cloudnative/kafka
tag: 3.6.0-debian-11-r0
listeners:
client:
containerPort: 9092
# 默许是带鉴权的,SASL_PLAINTEXT
protocol: PLAINTEXT
name: CLIENT
sslClientAuth: ""
controller:
replicaCount: 3 # 控制器的数量
persistence:
storageClass: "kafka-controller-local-storage"
size: "10Gi"
# 目录需求提早在宿主机上创立
local:
- name: kafka-controller-0
host: "local-168-182-110"
path: "/opt/bigdata/servers/kraft/kafka-controller/data1"
- name: kafka-controller-1
host: "local-168-182-111"
path: "/opt/bigdata/servers/kraft/kafka-controller/data1"
- name: kafka-controller-2
host: "local-168-182-112"
path: "/opt/bigdata/servers/kraft/kafka-controller/data1"
broker:
replicaCount: 3 # 署理的数量
persistence:
storageClass: "kafka-broker-local-storage"
size: "10Gi"
# 目录需求提早在宿主机上创立
local:
- name: kafka-broker-0
host: "local-168-182-110"
path: "/opt/bigdata/servers/kraft/kafka-broker/data1"
- name: kafka-broker-1
host: "local-168-182-111"
path: "/opt/bigdata/servers/kraft/kafka-broker/data1"
- name: kafka-broker-2
host: "local-168-182-112"
path: "/opt/bigdata/servers/kraft/kafka-broker/data1"
service:
type: NodePort
nodePorts:
#NodePort 默许范围是 30000-32767
client: "32181"
tls: "32182"
# Enable Prometheus to access ZooKeeper metrics endpoint
metrics:
enabled: true
kraft:
enabled: true
增加以下几个文件:
- kafka/templates/broker/pv.yaml
{{- range .Values.broker.persistence.local }}
---
apiVersion: v1
kind: PersistentVolume
metadata:
name: {{ .name }}
labels:
name: {{ .name }}
spec:
storageClassName: {{ $.Values.broker.persistence.storageClass }}
capacity:
storage: {{ $.Values.broker.persistence.size }}
accessModes:
- ReadWriteOnce
local:
path: {{ .path }}
nodeAffinity:
required:
nodeSelectorTerms:
- matchExpressions:
- key: kubernetes.io/hostname
operator: In
values:
- {{ .host }}
---
{{- end }}
- kafka/templates/broker/storage-class.yaml
kind: StorageClass
apiVersion: storage.k8s.io/v1
metadata:
name: {{ .Values.broker.persistence.storageClass }}
provisioner: kubernetes.io/no-provisioner
- kafka/templates/controller-eligible/pv.yaml
{{- range .Values.controller.persistence.local }}
---
apiVersion: v1
kind: PersistentVolume
metadata:
name: {{ .name }}
labels:
name: {{ .name }}
spec:
storageClassName: {{ $.Values.controller.persistence.storageClass }}
capacity:
storage: {{ $.Values.controller.persistence.size }}
accessModes:
- ReadWriteOnce
local:
path: {{ .path }}
nodeAffinity:
required:
nodeSelectorTerms:
- matchExpressions:
- key: kubernetes.io/hostname
operator: In
values:
- {{ .host }}
---
{{- end }}
- kafka/templates/controller-eligible/storage-class.yaml
kind: StorageClass
apiVersion: storage.k8s.io/v1
metadata:
name: {{ .Values.controller.persistence.storageClass }}
provisioner: kubernetes.io/no-provisioner
4)运用 Helm 布置 Kafka 集群
# 先预备好镜像
docker pull docker.io/bitnami/kafka:3.6.0-debian-11-r0
docker tag docker.io/bitnami/kafka:3.6.0-debian-11-r0 registry.cn-hangzhou.aliyuncs.com/bigdata_cloudnative/kafka:3.6.0-debian-11-r0
docker push registry.cn-hangzhou.aliyuncs.com/bigdata_cloudnative/kafka:3.6.0-debian-11-r0
# 开始装置
helm install kraft ./kafka -n kraft --create-namespace
NOTES
[root@local-168-182-110 KRaft-on-k8s]# helm upgrade kraft
Release "kraft" has been upgraded. Happy Helming!
NAME: kraft
LAST DEPLOYED: Sun Mar 24 20:05:04 2024
NAMESPACE: kraft
STATUS: deployed
REVISION: 3
TEST SUITE: None
NOTES:
CHART NAME: kafka
CHART VERSION: 26.0.0
APP VERSION: 3.6.0
** Please be patient while the chart is being deployed **
Kafka can be accessed by consumers via port 9092 on the following DNS name from within your cluster:
kraft-kafka.kraft.svc.cluster.local
Each Kafka broker can be accessed by producers via port 9092 on the following DNS name(s) from within your cluster:
kraft-kafka-controller-0.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092
kraft-kafka-controller-1.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092
kraft-kafka-controller-2.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092
kraft-kafka-broker-0.kraft-kafka-broker-headless.kraft.svc.cluster.local:9092
kraft-kafka-broker-1.kraft-kafka-broker-headless.kraft.svc.cluster.local:9092
kraft-kafka-broker-2.kraft-kafka-broker-headless.kraft.svc.cluster.local:9092
To create a pod that you can use as a Kafka client run the following commands:
kubectl run kraft-kafka-client --restart='Never' --image registry.cn-hangzhou.aliyuncs.com/bigdata_cloudnative/kafka:3.6.0-debian-11-r0 --namespace kraft --command -- sleep infinity
kubectl exec --tty -i kraft-kafka-client --namespace kraft -- bash
PRODUCER:
kafka-console-producer.sh
--broker-list kraft-kafka-controller-0.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092,kraft-kafka-controller-1.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092,kraft-kafka-controller-2.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092,kraft-kafka-broker-0.kraft-kafka-broker-headless.kraft.svc.cluster.local:9092,kraft-kafka-broker-1.kraft-kafka-broker-headless.kraft.svc.cluster.local:9092,kraft-kafka-broker-2.kraft-kafka-broker-headless.kraft.svc.cluster.local:9092
--topic test
CONSUMER:
kafka-console-consumer.sh
--bootstrap-server kraft-kafka.kraft.svc.cluster.local:9092
--topic test
--from-beginning
5)测验验证
# 创立客户端
kubectl run kraft-kafka-client --restart='Never' --image registry.cn-hangzhou.aliyuncs.com/bigdata_cloudnative/kafka:3.6.0-debian-11-r0 --namespace kraft --command -- sleep infinity
创立客户端
kafka-topics.sh --create --topic test --bootstrap-server kraft-kafka-controller-0.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092 --partitions 3 --replication-factor 2
# 查看概况
kafka-topics.sh --describe --bootstrap-server kraft-kafka-controller-0.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092 --topic test
# 删去topic
kafka-topics.sh --delete --topic test --bootstrap-server kraft-kafka-controller-0.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092
生产者和消费者
# 生产者
kafka-console-producer.sh
--broker-list kraft-kafka-controller-0.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092,kraft-kafka-controller-1.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092,kraft-kafka-controller-2.kraft-kafka-controller-headless.kraft.svc.cluster.local:9092,kraft-kafka-broker-0.kraft-kafka-broker-headless.kraft.svc.cluster.local:9092,kraft-kafka-broker-1.kraft-kafka-broker-headless.kraft.svc.cluster.local:9092,kraft-kafka-broker-2.kraft-kafka-broker-headless.kraft.svc.cluster.local:9092
--topic test
# 消费者
kafka-console-consumer.sh
--bootstrap-server kraft-kafka.kraft.svc.cluster.local:9092
--topic test
--from-beginning
6)更新集群
helm upgrade kraft ./kafka -n kraft
7)删去集群
helm uninstall kraft -n kraft
Raft Kafka on k8s 布置实战操作就先到这儿了,有任何疑问也可关注我公众号:大数据与云原生技能分享,进行技能交流,如本篇文章对您有所协助,费事帮忙一键三连(点赞、转发、保藏)~