StatefulSet在ZooKeeper和Kafka的实践

K8s的版本是1.7.6

采用nfs的nas存储模式

NFS的问题

• 建立zk集群的时候总是发现myid绑定一个id,先describe pod确认每个绑定不同的pvc,然后就确认是pv创建的问题，pv创建不能直接挂在一个大的存储上面，因为大家最后的目录相同/var/lib/zookeeper/data目录，所以无论哪个pvc挂上去都是同样的目录,解决办法，建立不同的存储挂载目录，然后分别挂载pv
• 建立pv的时候，指明storageClassName,比如

apiVersion: v1
kind: PersistentVolume
metadata:
    name: pv0003
spec:
    capacity:
      storage: 5Gi
    accessModes:
      - ReadWriteOnce
    persistentVolumeReclaimPolicy: Recycle
    storageClassName: nas-zk
    nfs:
      path: /k8s/weblogic
      server: 192.168.0.103

在使用pvc的时候，也指明storageClassName,比如

 volumeClaimTemplates:
  - metadata:
      name: datadir
    spec:
      accessModes: [ "ReadWriteOnce" ]
      resources:
        requests:
          storage: 10Gi
      storageClassName: nas-zk

这样就可以控制zk的存储使用的是带这个标签的pv

• pv和pvc的accessModes一定要保持一致，否则找不到

建立zk集群脚本

https://github.com/kubernetes/contrib/blob/master/statefulsets/zookeeper/zookeeper.yaml

也可以参考

https://kubernetes.io/docs/tutorials/stateful-application/zookeeper/

但是我的机器版本是1.7.6运行起来总是有问题,居然都不是一个个建立起来，而是一次性把所有的都建立起来。

改用脚本和镜像后问题才消失。

zk集群验证

for i in   ; do kubectl exec zk-$i -- hostname; done

zk-
zk-
zk-

for i in   ; do echo "myid zk-$i";kubectl exec zk-$i -- cat /var/lib/zookeeper/data/myid; done

myid zk-

myid zk-

myid zk-

for i in   ; do kubectl exec zk-$i -- hostname -f; done
zk-.zk-hs.default.svc.cluster.local
zk-.zk-hs.default.svc.cluster.local
zk-.zk-hs.default.svc.cluster.local

暴露服务给集群外

kubectl label pod zk- zkInst=
kubectl label pod zk- zkInst=
kubectl label pod zk- zkInst=

kubectl expose po zk- --port= --target-port= --name=zk- --selector=zkInst= --type=NodePort
kubectl expose po zk- --port= --target-port= --name=zk- --selector=zkInst= --type=NodePort
kubectl expose po zk- --port= --target-port= --name=zk-2 --selector=zkInst=2 --type=NodePort

建立kafka集群

构建脚本

https://github.com/kubernetes/contrib/blob/master/statefulsets/kafka/kafka.yaml

验证

root@kafka-:/opt/kafka/config# kafka-topics.sh --create \
> --topic test \
> --zookeeper zoo-.zk.default.svc.cluster.local:,zoo-.zk.default.svc.cluster.local:,zoo-.zk.default.svc.cluster.local: \
> --partitions  \
> --replication-factor 

Created topic "test".

root@kafka-:/opt/kafka/config# kafka-console-consumer.sh --topic test --bootstrap-server localhost:

root@kafka-:/# kafka-console-producer.sh --topic test --broker-list localhost:
I like kafka
hello world

#在消费者侧显示为：
I like kafka
hello world

参考

https://cloud.tencent.com/developer/article/1005492

zk.yaml

---
apiVersion: v1
kind: Service
metadata:
  name: zk-svc
  labels:
    app: zk-svc
spec:
  ports:
  - port:
    name: server
  - port:
    name: leader-election
  clusterIP: None
  selector:
    app: zk
---
apiVersion: v1
kind: ConfigMap
metadata:
  name: zk-cm
data:
  jvm.heap: "1G"
  tick: ""
  init: ""
  sync: ""
  client.cnxns: ""
  snap.retain: ""
  purge.interval: ""
---
apiVersion: policy/v1beta1
kind: PodDisruptionBudget
metadata:
  name: zk-pdb
spec:
  selector:
    matchLabels:
      app: zk
  minAvailable:
---
apiVersion: apps/v1beta1
kind: StatefulSet
metadata:
  name: zk
spec:
  serviceName: zk-svc
  replicas:
  template:
    metadata:
      labels:
        app: zk
    spec:
      affinity:
        podAntiAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            - labelSelector:
                matchExpressions:
                  - key: "app"
                    operator: In
                    values:
                    - zk
              topologyKey: "kubernetes.io/hostname"
      containers:
      - name: k8szk
        imagePullPolicy: Always
        image: gcr.io/google_samples/k8szk:v3
        resources:
          requests:
            memory: "2Gi"
            cpu: "500m"
        ports:
        - containerPort:
          name: client
        - containerPort:
          name: server
        - containerPort:
          name: leader-election
        env:
        - name : ZK_REPLICAS
          value: ""
        - name : ZK_HEAP_SIZE
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: jvm.heap
        - name : ZK_TICK_TIME
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: tick
        - name : ZK_INIT_LIMIT
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: init
        - name : ZK_SYNC_LIMIT
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: tick
        - name : ZK_MAX_CLIENT_CNXNS
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: client.cnxns
        - name: ZK_SNAP_RETAIN_COUNT
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: snap.retain
        - name: ZK_PURGE_INTERVAL
          valueFrom:
            configMapKeyRef:
                name: zk-cm
                key: purge.interval
        - name: ZK_CLIENT_PORT
          value: ""
        - name: ZK_SERVER_PORT
          value: ""
        - name: ZK_ELECTION_PORT
          value: ""
        command:
        - sh
        - -c
        - zkGenConfig.sh && zkServer.sh start-foreground
        readinessProbe:
          exec:
            command:
            - "zkOk.sh"
          initialDelaySeconds:
          timeoutSeconds:
        livenessProbe:
          exec:
            command:
            - "zkOk.sh"
          initialDelaySeconds:
          timeoutSeconds:
        volumeMounts:
        - name: datadir
          mountPath: /var/lib/zookeeper
      securityContext:
        runAsUser:
        fsGroup:
  volumeClaimTemplates:
  - metadata:
      name: datadir
    spec:
      accessModes: [ "ReadWriteOnce" ]
      resources:
        requests:
          storage: 10Gi

kafka.yaml

---
apiVersion: v1
kind: Service
metadata:
  name: kafka-svc
  labels:
    app: kafka
spec:
  ports:
  - port:
    name: server
  clusterIP: None
  selector:
    app: kafka
---
apiVersion: policy/v1beta1
kind: PodDisruptionBudget
metadata:
  name: kafka-pdb
spec:
  selector:
    matchLabels:
      app: kafka
  minAvailable:
---
apiVersion: apps/v1beta1
kind: StatefulSet
metadata:
  name: kafka
spec:
  serviceName: kafka-svc
  replicas:
  template:
    metadata:
      labels:
        app: kafka
    spec:
      affinity:
        podAntiAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            - labelSelector:
                matchExpressions:
                  - key: "app"
                    operator: In
                    values:
                    - kafka
              topologyKey: "kubernetes.io/hostname"
        podAffinity:
          preferredDuringSchedulingIgnoredDuringExecution:
             - weight:
               podAffinityTerm:
                 labelSelector:
                    matchExpressions:
                      - key: "app"
                        operator: In
                        values:
                        - zk
                 topologyKey: "kubernetes.io/hostname"
      terminationGracePeriodSeconds:
      containers:
      - name: k8skafka
        imagePullPolicy: Always
        image: gcr.io/google_samples/k8skafka:v1
        resources:
          requests:
            memory: "1Gi"
            cpu: 500m
        ports:
        - containerPort:
          name: server
        command:
        - sh
        - -c
        - "exec kafka-server-start.sh /opt/kafka/config/server.properties --override broker.id=${HOSTNAME##*-} \
          --override listeners=PLAINTEXT://:9093 \
          --override zookeeper.connect=zk-.zk-svc.default.svc.cluster.local:,zk-.zk-svc.default.svc.cluster.local:,zk-.zk-svc.default.svc.cluster.local: \
          --override log.dir=/var/lib/kafka \
          --override auto.create.topics.enable=true \
          --override auto.leader.rebalance.enable=true \
          --override background.threads= \
          --override compression.type=producer \
          --override delete.topic.enable=false \
          --override leader.imbalance.check.interval.seconds= \
          --override leader.imbalance.per.broker.percentage= \
          --override log.flush.interval.messages= \
          --override log.flush.offset.checkpoint.interval.ms= \
          --override log.flush.scheduler.interval.ms= \
          --override log.retention.bytes=- \
          --override log.retention.hours= \
          --override log.roll.hours= \
          --override log.roll.jitter.hours= \
          --override log.segment.bytes= \
          --override log.segment.delete.delay.ms= \
          --override message.max.bytes= \
          --override min.insync.replicas= \
          --override num.io.threads= \
          --override num.network.threads= \
          --override num.recovery.threads.per.data.dir= \
          --override num.replica.fetchers= \
          --override offset.metadata.max.bytes= \
          --override offsets.commit.required.acks=- \
          --override offsets.commit.timeout.ms= \
          --override offsets.load.buffer.size= \
          --override offsets.retention.check.interval.ms= \
          --override offsets.retention.minutes= \
          --override offsets.topic.compression.codec= \
          --override offsets.topic.num.partitions= \
          --override offsets.topic.replication.factor= \
          --override offsets.topic.segment.bytes= \
          --override queued.max.requests= \
          --override quota.consumer.default= \
          --override quota.producer.default= \
          --override replica.fetch.min.bytes= \
          --override replica.fetch.wait.max.ms= \
          --override replica.high.watermark.checkpoint.interval.ms= \
          --override replica.lag.time.max.ms= \
          --override replica.socket.receive.buffer.bytes= \
          --override replica.socket.timeout.ms= \
          --override request.timeout.ms= \
          --override socket.receive.buffer.bytes= \
          --override socket.request.max.bytes= \
          --override socket.send.buffer.bytes= \
          --override unclean.leader.election.enable=true \
          --override zookeeper.session.timeout.ms= \
          --override zookeeper.set.acl=false \
          --override broker.id.generation.enable=true \
          --override connections.max.idle.ms= \
          --override controlled.shutdown.enable=true \
          --override controlled.shutdown.max.retries= \
          --override controlled.shutdown.retry.backoff.ms= \
          --override controller.socket.timeout.ms= \
          --override default.replication.factor= \
          --override fetch.purgatory.purge.interval.requests= \
          --override group.max.session.timeout.ms= \
          --override group.min.session.timeout.ms= \
          --override inter.broker.protocol.version=0.10.-IV0 \
          --override log.cleaner.backoff.ms= \
          --override log.cleaner.dedupe.buffer.size= \
          --override log.cleaner.delete.retention.ms= \
          --override log.cleaner.enable=true \
          --override log.cleaner.io.buffer.load.factor=0.9 \
          --override log.cleaner.io.buffer.size= \
          --override log.cleaner.io.max.bytes.per.second=1.7976931348623157E308 \
          --override log.cleaner.min.cleanable.ratio=0.5 \
          --override log.cleaner.min.compaction.lag.ms= \
          --override log.cleaner.threads= \
          --override log.cleanup.policy=delete \
          --override log.index.interval.bytes= \
          --override log.index.size.max.bytes= \
          --override log.message.timestamp.difference.max.ms= \
          --override log.message.timestamp.type=CreateTime \
          --override log.preallocate=false \
          --override log.retention.check.interval.ms= \
          --override max.connections.per.ip= \
          --override num.partitions= \
          --override producer.purgatory.purge.interval.requests= \
          --override replica.fetch.backoff.ms= \
          --override replica.fetch.max.bytes= \
          --override replica.fetch.response.max.bytes= \
          --override reserved.broker.max.id= "
        env:
        - name: KAFKA_HEAP_OPTS
          value : "-Xmx512M -Xms512M"
        - name: KAFKA_OPTS
          value: "-Dlogging.level=INFO"
        volumeMounts:
        - name: datadir
          mountPath: /var/lib/kafka
        readinessProbe:
          exec:
           command:
            - sh
            - -c
            - "/opt/kafka/bin/kafka-broker-api-versions.sh --bootstrap-server=localhost:9093"
      securityContext:
        runAsUser:
        fsGroup:
  volumeClaimTemplates:
  - metadata:
      name: datadir
    spec:
      accessModes: [ "ReadWriteOnce" ]
      resources:
        requests:
          storage: 10Gi