Docker（二十一）k8s 部署分布式jenkins

前提条件是，有storageclass，利用pvc 创建持久化存储 创建kube-ops namespace 

这里创建opspvc 另外把accessmode 换成readwritemany，因为会有多个pod 进行读写

然后部署 jenkins master deployment 如下

---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: jenkins
  namespace: kube-ops
spec:
  template:
    metadata:
      labels:
        app: jenkins
    spec:
      terminationGracePeriodSeconds: 10
      serviceAccountName: jenkins
      containers:
      - name: jenkins
        image: jenkins/jenkins:lts
        imagePullPolicy: IfNotPresent
        ports:
        - containerPort: 8080
          name: web
          protocol: TCP
        - containerPort: 50000
          name: agent
          protocol: TCP
        resources:
          limits:
            cpu: 2000m
            memory: 4Gi
          requests:
            cpu: 1000m
            memory: 2Gi
        livenessProbe:
          httpGet:
            path: /login
            port: 8080
          initialDelaySeconds: 60
          timeoutSeconds: 5
          failureThreshold: 12
        readinessProbe:
          httpGet:
            path: /login
            port: 8080
          initialDelaySeconds: 60
          timeoutSeconds: 5
          failureThreshold: 12
        volumeMounts:
        - name: jenkinshome
          subPath: jenkins
          mountPath: /var/jenkins_home
        env:
        - name: LIMITS_MEMORY
          valueFrom:
            resourceFieldRef:
              resource: limits.memory
              divisor: 1Mi
        - name: JAVA_OPTS
          value: -Xmx$(LIMITS_MEMORY)m -XshowSettings:vm -Dhudson.slaves.NodeProvisioner.initialDelay=0 -Dhudson.slaves.NodeProvisioner.MARGIN=50 -Dhudson.slaves.NodeProvisioner.MARGIN0=0.85 -Duser.timezone=Asia/Shanghai
      securityContext:
        fsGroup: 1000
      volumes:
      - name: jenkinshome
        persistentVolumeClaim:
          claimName: opspvc

---
apiVersion: v1
kind: Service
metadata:
  name: jenkins
  namespace: kube-ops
  labels:
    app: jenkins
spec:
  selector:
    app: jenkins
  ports:
  - name: web
    port: 8080
    targetPort: web
  - name: agent
    port: 50000
    targetPort: agent

分配权限，配置rbac如下

apiVersion: v1
kind: ServiceAccount
metadata:
  name: jenkins
  namespace: kube-ops

---

kind: Role
apiVersion: rbac.authorization.k8s.io/v1beta1
metadata:
  name: jenkins
  namespace: kube-ops
rules:
  - apiGroups: [""]
    resources: ["pods"]
    verbs: ["create","delete","get","list","patch","update","watch"]
  - apiGroups: [""]
    resources: ["pods/exec"]
    verbs: ["create","delete","get","list","patch","update","watch"]
  - apiGroups: [""]
    resources: ["pods/log"]
    verbs: ["get","list","watch"]
  - apiGroups: [""]
    resources: ["secrets"]
    verbs: ["get"]

---
apiVersion: rbac.authorization.k8s.io/v1beta1
kind: RoleBinding
metadata:
  name: jenkins
  namespace: kube-ops
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: Role
  name: jenkins
subjects:
  - kind: ServiceAccount
    name: jenkins
    namespace: kube-ops

第二步，安装k8s 插件，与k8s 目标集群进行远程连接

因为是腾讯云，并没有提供tls 客户端认证，所以直接利用账号密码，进行认证，记得，禁用https 证书检查

jenkins和k8s 就集成好了

划重点！！！：jenkins地址，这里我这里写的是内网地址以及暴露了50000端口（用来与slave 建立通信使用），因为master 和slave 分别在不通的k8s 集群里，那么需要远程进行联通，而jenkins-ui 我是以ingress 的方式对外暴露

一下的slave 的配置

这里需要注意的是标签列表，这里填写的标签，需要在slave 所在k8s 集群的节点上进行标注，而这个名字，是label这个字段里的key 并未是value，这里要注意

之后再job 里配置

这个意思就是 slave 会尽可能的在这个节点build