• kubernetes 控制器详解【持续完善中】


    资源创建详解

    一:Pod及常用参数

    1.简介

    2.模板

    3.删除pod

    示例流程如下:

    1. 用户发送删除pod的命令,默认宽限期是30秒;
    2. 在Pod超过该宽限期后API server就会更新Pod的状态为“dead”;
    3. 在客户端命令行上显示的Pod状态为“terminating”;
    4. 跟第三步同时,当kubelet发现pod被标记为“terminating”状态时,开始停止pod进程:
    5. 如果在pod中定义了preStop hook,在停止pod前会被调用。如果在宽限期过后,preStop hook依然在运行,第二步会再增加2秒的宽限期;
    6. 向Pod中的进程发送TERM信号;
    7. 跟第三步同时,该Pod将从该service的端点列表中删除,不再是replication controller的一部分。关闭的慢的pod将继续处理load balancer转发的流量;
    8. 过了宽限期后,将向Pod中依然运行的进程发送SIGKILL信号而杀掉进程。
    9. Kublete会在API server中完成Pod的的删除,通过将优雅周期设置为0(立即删除)。Pod在API中消失,并且在客户端也不可见。

    3.1.默认删除

    默认删除,会按照上线的流程,等待宽限期30s

    kubectl delete POD --namespace=xxx
    

    3.2.强制删除

    设置宽限期为0,会立即删除,没有宽限期

    kubectl delete POD --namespace=xxx --force --grace-period=0
    

    4.设置Pod主机名

    template.spec.hostname:pod_name 设置pod的主机名

    5.镜像拉取策略(ImagePullPolicy)

    ImagePullPolicy:

    • Always: 不管镜像是否存在,都会拉取
    • Never:不管镜像是否存在,都不会拉取
    • IfNotPresent:只有当镜像不存在的时候才会进行拉取

    注意:

    • 默认为IfNotPresent,但:latest标签的镜像默认为Always
    • 拉取镜像时docker会进行校验,如果镜像中的MD5码没有变,则不会拉取镜像数据。
    • 生产环境中应该尽量避免使用:latest标签,而开发环境中可以借助:latest标签自动拉取最新的镜像。

    二:RC

    1.简介

    2.模板

    官网资料

    apiVersion: v1
    kind: ReplicationController	# 定义资源类型
    metadata:
      name: zabbix-db
      namespace: zabbix
    spec:
      replicas: 1
      selector:
        app: zabbix-db
      template:
        metadata:
          name: zabbix-db
          labels:
            app: zabbix-db
        spec:
          terminationGracePeriodSeconds: 30 # 容器平滑退出时间,默认30s
          hostname: zabbix-db # 设置容器的主机名 
          containers:
          - name: zabbix-db
            image: mysql:5.7.22
            env:
            - name: MYSQL_DATABASE
              value: zabbix
            - name: MYSQL_USER
              value: zabbix
            - name: MYSQL_PASSWORD
              value: zabbix
            - name: MYSQL_ROOT_PASSWORD
              value: Abc123@
            ports:
            - containerPort: 3306
            volumeMounts:
            - mountPath: /var/lib/mysql
              readOnly: false
              name: zabbix-database
          volumes:
          - name: zabbix-database
            nfs:
              server: 172.30.80.222
              path: "/data/zabbix/zabbix_db/mysql"
    

    三:Deployment

    1.简介

    ​ Deployment为Pod和Replica Set(升级版的 Replication Controller)提供声明式更新。比于RC,Deployment直接使用kubectl edit deployment/deploymentName 或者kubectl set方法就可以直接升级(原理是Pod的template发生变化,例如更新label、更新镜像版本等操作会触发Deployment的滚动升级)。

    2.模板

    官方资料

    apiVersion: extensions/v1beta1
    kind: Deployment
    metadata:
      name: nginx
      labels:
        app: nginx
    spec:
      minReadySeconds: 30	# 滚动升级时,容器准备就绪时间最少为30s
      replicas: 5
      selector:
        matchLabels:
          app: nginx
      strategy:
        rollingUpdate:	# 当type为RollingUpdate时,才会进行设置
          maxSurge: 25%	# 当定义为25%时,容器会先新建百分之25的pod,然后开始滚动升级
          maxUnavailable: 25%	# 每次升级的百分比,也可以是绝对数(5),默认值25%
        type: RollingUpdate	# 滚动升级方式,有Recreate:立马关闭所有pod进行升级,RollingUpdate:采用百分比的方式滚动升级
      template:		# 模板
        metadata:
          creationTimestamp: null
          labels:
            app: nginx
        spec:
          containers:
          - name: nginx
            image: nginx:1.7.9
            imagePullPolicy: IfNotPresent
            resources:	# 资源限制
              requests:
                memory: "64Mi"
                cpu: "250m"
              limits:
                memory: "128Mi"
                cpu: "500m"
            env:	# 设置变量
            - name: DB_SERVER_HOST
              value: zabbix-db-server
            - name: MYSQL_DATABASE
              value: zabbix
            - name: MYSQL_USER
              value: zabbix
            - name: MYSQL_PASSWORD
              value: zabbix
            - name: MYSQL_ROOT_PASSWORD
              value: Abc123@
            - name: ZBX_HISTORYSTORAGEURL
              value: http://192.168.2.171:9200
            - name: ZBX_HISTORYSTORAGETYPES
              value: uint,dbl,str,log,text
            ports:
            - containerPort: 80   
           	livenessProbe:	#livenessProbe是K8S认为该pod是存活的,不存在则需要kill掉,然后再新启动一个,以达到RS指定的个数。
              httpGet:
                path: /
                port: 80
                scheme: HTTP
              initialDelaySeconds: 30
              timeoutSeconds: 5
              successThreshold: 1
              failureThreshold: 5
            readinessProbe:	#readinessProbe是K8S认为该pod是启动成功的,这里根据每个应用的特性,自己去判断,可以执行command,也可以进行httpGet。
              httpGet:
                path: /
                port: 80
                scheme: HTTP
              initialDelaySeconds: 30
              timeoutSeconds: 5
              successThreshold: 1
            restartPolicy: Always	# 启动失败时,会重试启动
    
    ---
    apiVersion: v1
    kind: Service
    metadata:
      name: nginx-service
    spec:
      type: NodePort
      sessionAffinity: ClientIP
      selector:
        app: nginx
      ports:
        # 将容器的80端口映射到master主机的8888端口
        - port: 80	# pod端口
          nodePort: 8888	# 宿主机上的端口
    

    四:HPA

    1.简介

    ​ Horizontal Pod Autoscaler根据观察到的CPU利用率自动调整复制控制器,部署或副本集中的容器数量(或者,通过 自定义指标 支持,根据其他一些应用程序提供的指标)。请注意,Horizontal Pod Autoscaling不适用于无法缩放的对象,例如DaemonSet。

    2.模板

    创建一个pod,必须添加资源请求和限制参数

    apiVersion: extensions/v1beta1
    kind: Deployment
    metadata:
      name: nginx
      labels:
        app: nginx
    spec:
      minReadySeconds: 30
      replicas: 5
      selector:
        matchLabels:
          app: nginx
      strategy:
        rollingUpdate:
          maxSurge: 25%
          maxUnavailable: 25%
        type: RollingUpdate
      template:
        metadata:
          creationTimestamp: null
          labels:
            app: nginx
        spec:
          containers:
          - name: nginx
            image: nginx:1.7.9
            imagePullPolicy: IfNotPresent
            resources:	
              requests:		# 在使用HPA自动扩展时,必须使用资源请求和资源限制
                memory: "64Mi"
                cpu: "250m"
              limits:
                memory: "128Mi"
                cpu: "500m"
            ports:
            - containerPort: 80
          restartPolicy: Always
    
    ---
    apiVersion: v1
    kind: Service
    metadata:
      name: nginx-service
    spec:
      type: NodePort
      sessionAffinity: ClientIP
      selector:
        app: nginx
      ports:
        # 将容器的80端口映射到master主机的8888端口
        - port: 80
          nodePort: 8888
    

    创建HPA

    V2 HPA创建规则

    官方资料

    apiVersion: autoscaling/v1
    kind: HorizontalPodAutoscaler	# 指定对象类型
    metadata:
      name: nginx-hpa	# 名字
      labels:	# 标签
        app: hpa	
        version: v0.0.1
    spec:
      scaleTargetRef:
        apiVersion: v1
        kind: Deployment
        name: nginx	# 创建Deployment时,指定的名字
      minReplicas: 1	# 最小pod
      maxReplicas: 10	# 最大pod
      targetCPUUtilizationPercentage: 70	# CPU用到70%,自动扩展一个pod
    

    查看

    # kubectl get hpa
    NAME        REFERENCE          TARGETS    MINPODS   MAXPODS   REPLICAS   AGE
    nginx-hpa   Deployment/nginx   0% / 70%   1         10        1          4d
    

    查看详情

    # kubectl describe hpa/nginx-hpa
    Name:                                                  nginx-hpa
    Namespace:                                             default
    Labels:                                                app=hpa
                                                           version=v0.0.1
    Annotations:                                           <none>
    CreationTimestamp:                                     Thu, 02 Aug 2018 16:39:55 +0800
    Reference:                                             Deployment/nginx
    Metrics:                                               ( current / target )
      resource cpu on pods  (as a percentage of request):  0% (0) / 70%
    Min replicas:                                          1
    Max replicas:                                          10
    Conditions:
      Type            Status  Reason            Message
      ----            ------  ------            -------
      AbleToScale     True    ReadyForNewScale  the last scale time was sufficiently old as to warrant a new scale
      ScalingActive   True    ValidMetricFound  the HPA was able to succesfully calculate a replica count from cpu resource utilization (percentage of request)
      ScalingLimited  True    TooFewReplicas    the desired replica count was less than the minimum replica count
    Events:           <none>
    
    

    五:StatefulSet

    1.简介

    2.模板

    2.1.基础环境介绍

    StatefulSet Name Service Name
    apiVersion: v1
    kind: Service
    metadata:
      name: zoo01
      labels:
        app: zoo01
    spec:
      ports:
      - port: 2888
        name: leader-listen
      - port: 3888
        name: leader-vote
      clusterIP: None
      selector:
        app: zoo01
    
    ---
    apiVersion: apps/v1beta1
    kind: StatefulSet
    metadata:
      name: zoo01
      labels:
        app: zoo01
    spec:
      serviceName: "zoo01-service"
      replicas: 1
      template:
        metadata:
          creationTimestamp: null
          labels:
            app: zoo01
        spec:
          terminationGracePeriodSeconds: 30
          hostname: zoo01
          containers:
          - name: zoo01
            image: zookeeper:3.5
            imagePullPolicy: IfNotPresent
            resources:
              requests:
                memory: "64Mi"
                cpu: "250m"
              limits:
                memory: "128Mi"
                cpu: "500m"
            env:
              - name: ZOO_MY_ID
                value: "1"
              - name: ZOO_SERVERS
                value: "server.1=zoo01:2888:3888 server.2=zoo02:2888:3888 server.3=zoo03:2888:3888"
            ports:
              - containerPort: 2181
              - containerPort: 2888
              - containerPort: 3888
          restartPolicy: Always
    

    六:PV和PVC

    PV

    apiVersion: v1
    kind: PersistentVolume
    metadata:
      name: gitlab-pv
      namespace: dev
    spec:
      capacity:
        storage: 10Gi	# 大小
      accessModes:
        - ReadWriteOnce	# 读写模式
      persistentVolumeReclaimPolicy: Recycle
      storageClassName: slow
      mountOptions:	# 挂载选项
        - hard
        - nfsvers=4.1
      nfs:	# nfs服务器地址
        path: /data/gitlab
        server: 172.30.80.222
    

    PVC

    kind: PersistentVolumeClaim
    apiVersion: v1
    metadata:
      name: gitlab-pvc
      namespace: dev
      labels:
        type: nfs
    spec:
      accessModes:
        - ReadWriteOnce	# PV和PVC的绑定关系,主要依靠读写模式和存储大小
      storageClassName: slow
      resources:
        requests:
          storage: 10Gi	 
      selector:
        matchLabels:
          name: gitlab-pv
    

    Pod

    
    

    八:扩展

    8.1.Pod调度到指定的Node

    介绍:

    ​ Pod.spec.nodeSelector通过kubernetes的label-selector机制选择节点,由调度器调度策略匹配label,而后调度Pod到目标节点,该匹配规则属于强制约束。

    apiVersion: extensions/v1beta1
    kind: Deployment
    metadata:
      name: zabbix-server
      namespace: zabbix
    spec:
      replicas: 1
      selector:
        matchLabels:
          app: zabbix-server
      template:
        metadata:
          name: zabbix-server
          labels:
            app: zabbix-server
        spec:
          nodeSelector:	# 基于node的lable进行调度
            kubernetes.io/hostname: 172.30.80.220	# 指定主机的lable标签
          containers:
          - name: zabbix-server
            image: zabbix/zabbix-server-mysql:latest
            env:
            - name: DB_SERVER_HOST
              value: zabbix-db-server
            - name: MYSQL_DATABASE
              value: zabbix
            - name: MYSQL_USER
              value: zabbix
            - name: MYSQL_PASSWORD
              value: zabbix
            - name: MYSQL_ROOT_PASSWORD
              value: Abc123@
            - name: ZBX_HISTORYSTORAGEURL
              value: http://192.168.2.171:9200
            - name: ZBX_HISTORYSTORAGETYPES
              value: uint,dbl,str,log,text
            ports:
            - containerPort: 10051
    
    
  • 相关阅读:
    我的五年百度博客文章列表
    servlet过滤器2 解决用户非法在线 filter
    flexpaper实例在浏览器无法显示的解决办法
    我的五年百度博客文章列表(带链接版)
    servlet过滤器1 解决字符集乱码 filter
    GBK和UTF8之间的战争,websphere6.1乱码解决方案
    ORACLE SQL性能优化系列 (一)
    ORACLE SQL性能优化系列 (四)
    Oracle优化SQL查询优化研究
    用SQL*Loader将Excel数据导出到Oracle
  • 原文地址:https://www.cnblogs.com/GXLo/p/9983716.html
Copyright © 2020-2023  润新知