跳到主要内容

05、Kubernetes - 实战:pod控制器

一、Pod控制器

Pod 的分类

自主式Pod:Pod 退出后不会被创建
控制器管理的 Pod:在控制器的生命周期里,始终要维持 Pod 的副本数目

控制器类型

Replication Controller和ReplicaSet
Deployment
DaemonSet
StatefulSet
Job
CronJob
HPA全称Horizontal Pod Autoscaler

Replication Controller和ReplicaSet

ReplicaSet (RS)是下一代的 Replication Controller(RC),官方推荐使用ReplicaSet。

ReplicaSet 和 Replication Controller 的唯一区别是选择器的支持,ReplicaSet 支持新的基于集合的选择器需求。

ReplicaSet 确保任何时间都有指定数量的 Pod 副本在运行。

虽然ReplicaSets 可以独立使用,但今天它主要被Deployments 用作协调 Pod 创建、删除和更新的机制。

Deployment

Deployment 为 Pod 和 ReplicaSet 提供了一个申明式的定义方法。
典型的应用场景:

用来创建Pod和ReplicaSet
滚动更新和回滚
扩容和缩容
暂停与恢复

DaemonSet

DaemonSet 确保全部(或者某些)节点上运行一个 Pod 的副本。当有节点加入集群时, 也会为他们新增一个 Pod 。当有节点从集群移除时,这些 Pod 也会被回收。删除 DaemonSet 将会删除它创建的所有 Pod。

DaemonSet 的典型用法:

在每个节点上运行集群存储 DaemonSet,例如 glusterd、ceph。
在每个节点上运行日志收集 DaemonSet,例如 fluentd、logstash。
在每个节点上运行监控 DaemonSet,例如 Prometheus Node Exporter、zabbix agent等

一个简单的用法是在所有的节点上都启动一个 DaemonSet,将被作为每种类型的 daemon 使用。

一个稍微复杂的用法是单独对每种 daemon 类型使用多个 DaemonSet,但具有不同的标志, 并且对不同硬件类型具有不同的内存、CPU 要求。

StatefulSet

StatefulSet 是用来管理有状态应用的工作负载 API 对象。实例之间有不对等关系,以及实例对外部数据有依赖关系的应用,称为“有状态应用”

StatefulSet 用来管理 Deployment 和扩展一组 Pod,并且能为这些 Pod 提供序号和唯一性保证。

StatefulSets 对于需要满足以下一个或多个需求的应用程序很有价值:

稳定的、唯一的网络标识符。
稳定的、持久的存储。
有序的、优雅的部署和缩放。
有序的、自动的滚动更新。

Job

执行批处理任务,仅执行一次任务,保证任务的一个或多个Pod成功结束。

CronJob

Cron Job 创建基于时间调度的 Jobs。

一个CronJob 对象就像 crontab (cron table) 文件中的一行,它用 Cron 格式进行编写,并周期性地在给定的调度时间执行 Job。
HPA

根据资源利用率自动调整service中Pod数量,实现Pod水平自动缩放。

二、Pod控制器使用示例

ReplicaSet举例

编辑以下yaml文件:

[root@server1 ~]# vim rs.yaml 
[root@server1 ~]# cat rs.yaml 
apiVersion: apps/v1
kind: ReplicaSet
metadata:
    name: replicaset-example
spec:
  replicas: 2			#启动pod的副本数
  selector:				#定义选择器为标签选择器。
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx				#定义容器的标签
    spec:						#定义容器
      containers:
      - name: nginx
        image: nginx

以上使用的控制器的RS,RS控制器通过pod的标签(matchLabels)来控制pod的数量,使用apply命令可以实现创建,更新pod,而create命令创建后若需要更新只能删除之后再创建,因此建议使用apply:

[root@server1 ~]# kubectl apply -f rs.yaml 
replicaset.apps/replicaset-example created
[root@server1 ~]# kubectl apply -f rs.yaml 
replicaset.apps/replicaset-example unchanged

查看状态:

[root@server1 ~]# kubectl get pod
NAME                       READY   STATUS    RESTARTS   AGE
replicaset-example-p6sv6   1/1     Running   0          34s
replicaset-example-s5jps   1/1     Running   0          34s
[root@server1 ~]# kubectl get rs			#获取rs的信息
NAME                 DESIRED   CURRENT   READY   AGE
replicaset-example   2         2         2       36s
[root@server1 ~]# kubectl get pod -o wide
NAME                       READY   STATUS    RESTARTS   AGE   IP            NODE      NOMINATED NODE   READINESS GATES
replicaset-example-p6sv6   1/1     Running   0          38s   10.244.1.14   server2   <none>           <none>
replicaset-example-s5jps   1/1     Running   0          38s   10.244.2.21   server3   <none>           <none>

可以看出一且正常,接下来进行pod的拉伸与压缩:

拉伸:

[root@server1 ~]# vim rs.yaml 
[root@server1 ~]# cat rs.yaml 
apiVersion: apps/v1
kind: ReplicaSet
metadata:
    name: replicaset-example
spec:
  replicas: 4		#拉伸为4个
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx

[root@server1 ~]# kubectl apply -f rs.yaml 
replicaset.apps/replicaset-example configured

查看pod数:

[root@server1 ~]# kubectl get pod
NAME                       READY   STATUS    RESTARTS   AGE
replicaset-example-p6sv6   1/1     Running   0          97s
replicaset-example-s5jps   1/1     Running   0          97s
replicaset-example-wzclz   1/1     Running   0          21s
replicaset-example-zk4mb   1/1     Running   0          21s

已经被拉伸成了4个。

缩减:

[root@server1 ~]# vim rs.yaml 
[root@server1 ~]# cat rs.yaml 
apiVersion: apps/v1
kind: ReplicaSet
metadata:
    name: replicaset-example
spec:
  replicas: 2			#缩减为2个
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx

[root@server1 ~]# kubectl apply -f rs.yaml 
replicaset.apps/replicaset-example configured
[root@server1 ~]# kubectl get pod
NAME                       READY   STATUS        RESTARTS   AGE
replicaset-example-p6sv6   1/1     Running       0          2m6s
replicaset-example-s5jps   1/1     Running       0          2m6s

可以看出删除的是刚刚创建的那两个pod。

接下来更改一个pod 的标签:

[root@server1 ~]# kubectl label pod replicaset-example-s5jps app=myapp --overwrite 		#将标签强制更改为myapp
pod/replicaset-example-s5jps labeled
[root@server1 ~]# kubectl get pod --show-labels
NAME                       READY   STATUS    RESTARTS   AGE     LABELS
replicaset-example-p6sv6   1/1     Running   0          3m56s   app=nginx
replicaset-example-s5jps   1/1     Running   0          3m56s   app=myapp
replicaset-example-w98nk   1/1     Running   0          26s     app=nginx

可以看出现在有3个pod,两个标签为nginx一个为myapp,RS控制器的工作原理就是维持标签为nginx的pod个数有2个,因此当我们更改一个pod的标签后,RS又会给我们创建一个标签为nginx的pod,而当我们将标签为myapp的pod删除后RS控制器不会有操作:

[root@server1 ~]# kubectl delete pod replicaset-example-s5jps
pod "replicaset-example-s5jps" deleted
[root@server1 ~]# kubectl get pod --show-labels
NAME                       READY   STATUS    RESTARTS   AGE    LABELS
replicaset-example-p6sv6   1/1     Running   0          5m8s   app=nginx
replicaset-example-w98nk   1/1     Running   0          98s    app=nginx

此时我们再更改:

[root@server1 ~]# kubectl label pod replicaset-example-p6sv6 app=myapp --overwrite 		#首先保证3个pod
pod/replicaset-example-p6sv6 labeled
[root@server1 ~]# kubectl get pod --show-labels
NAME                       READY   STATUS    RESTARTS   AGE     LABELS
replicaset-example-p6sv6   1/1     Running   0          6m4s    app=myapp
replicaset-example-w98nk   1/1     Running   0          2m34s   app=nginx
replicaset-example-x2lq9   1/1     Running   0          26s     app=nginx
[root@server1 ~]# kubectl label pod replicaset-example-p6sv6 app=nginx --overwrite
pod/replicaset-example-p6sv6 labeled
[root@server1 ~]# kubectl get pod --show-labels
NAME                       READY   STATUS        RESTARTS   AGE     LABELS
replicaset-example-p6sv6   1/1     Running       0          6m23s   app=nginx
replicaset-example-w98nk   1/1     Running       0          2m53s   app=nginx
replicaset-example-x2lq9   0/1     Terminating   0          45s     app=nginx
[root@server1 ~]# kubectl get pod --show-labels
NAME                       READY   STATUS    RESTARTS   AGE     LABELS
replicaset-example-p6sv6   1/1     Running   0          6m25s   app=nginx
replicaset-example-w98nk   1/1     Running   0          2m55s   app=nginx

以上实验可以看出,当集群里有3个标签是nginx的pod的时候,RS控制器又会帮我们将最后创建的pod删除。

实验后删除:

[root@server1 ~]# kubectl delete -f rs.yaml 
replicaset.apps "replicaset-example" deleted

Deployment控制器示例

编辑yaml文件:

[root@server1 ~]# vim deployment.yaml 
[root@server1 ~]# cat deployment.yaml 
apiVersion: apps/v1
kind: Deployment
metadata:
  name: deployment-nginx
  labels:
    app: nginx
spec:
  replicas: 2
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: ikubernetes/myapp:v1
        ports:
        - containerPort: 80

创建pod:

[root@server1 ~]# kubectl apply -f deployment.yaml 
deployment.apps/deployment-nginx created
[root@server1 ~]# kubectl get pod
NAME                                READY   STATUS    RESTARTS   AGE
deployment-nginx-56d786cd98-kx5pw   1/1     Running   0          23s
deployment-nginx-56d786cd98-qgcjl   1/1     Running   0          23s

[root@server1 ~]# kubectl get rs
NAME                          DESIRED   CURRENT   READY   AGE
deployment-nginx-56d786cd98   2         2         2       28s
[root@server1 ~]# kubectl get deployments.apps 		#查看deployments
NAME               READY   UP-TO-DATE   AVAILABLE   AGE
deployment-nginx   2/2     2            2           32s

以上运行结果可以看出deployments底层也是由RS实现的,接下来进行拉伸:

[root@server1 ~]# vim deployment.yaml 
[root@server1 ~]# cat deployment.yaml 
apiVersion: apps/v1
kind: Deployment
metadata:
  name: deployment-nginx
  labels:
    app: nginx
spec:
  replicas: 4			#拉伸为4个
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: ikubernetes/myapp:v1
        ports:
        - containerPort: 80

[root@server1 ~]# kubectl apply -f deployment.yaml 
deployment.apps/deployment-nginx configured
[root@server1 ~]# kubectl get pod
NAME                                READY   STATUS    RESTARTS   AGE
deployment-nginx-56d786cd98-942dh   1/1     Running   0          26s
deployment-nginx-56d786cd98-kx5pw   1/1     Running   0          108s
deployment-nginx-56d786cd98-qgcjl   1/1     Running   0          108s
deployment-nginx-56d786cd98-zb8s8   1/1     Running   0          26s

可以看出已经拉伸为4个。接下来进行滚动更新:

[root@server1 ~]# vim deployment.yaml 
[root@server1 ~]# cat deployment.yaml 
apiVersion: apps/v1
kind: Deployment
metadata:
  name: deployment-nginx
  labels:
    app: nginx
spec:
  replicas: 4
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: ikubernetes/myapp:v2		#更新到v2
        ports:
        - containerPort: 80

[root@server1 ~]# kubectl apply -f deployment.yaml 
deployment.apps/deployment-nginx configured
[root@server1 ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE     IP            NODE      NOMINATED NODE   READINESS GATES
deployment-nginx-868855d887-2zh45   1/1     Running   0          2m39s   10.244.2.26   server3   <none>           <none>
deployment-nginx-868855d887-87m22   1/1     Running   0          2m34s   10.244.1.20   server2   <none>           <none>
deployment-nginx-868855d887-hb6mv   1/1     Running   0          2m39s   10.244.1.19   server2   <none>           <none>
deployment-nginx-868855d887-v8ndt   1/1     Running   0          2m33s   10.244.2.27   server3   <none>           <none>
[root@server1 ~]# curl 10.244.2.26
Hello MyApp | Version: v2 | <a href="hostname.html">Pod Name</a>

可以看出版本已经更新到了v2,我们查看rs可以看出更新过程:

[root@server1 ~]# kubectl get rs
NAME                          DESIRED   CURRENT   READY   AGE
deployment-nginx-56d786cd98   0         0         0       4m41s
deployment-nginx-868855d887   4         4         4       2m27s

可以看出更新时控制器新建一个rs,然后再新建4个pod,原来的rs依然存在,为了方便我们进行回滚:

[root@server1 ~]# vim deployment.yaml 
[root@server1 ~]# cat deployment.yaml 
apiVersion: apps/v1
kind: Deployment
metadata:
  name: deployment-nginx
  labels:
    app: nginx
spec:
  replicas: 4
  selector:
    matchLabels:
      app: nginx
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: ikubernetes/myapp:v1
        ports:
        - containerPort: 80

[root@server1 ~]# kubectl apply -f deployment.yaml 
deployment.apps/deployment-nginx configured
[root@server1 ~]# kubectl get pod
NAME                                READY   STATUS              RESTARTS   AGE
deployment-nginx-56d786cd98-78m7z   0/1     ContainerCreating   0          4s
deployment-nginx-56d786cd98-8hnns   1/1     Running             0          9s
deployment-nginx-56d786cd98-lcvzw   1/1     Running             0          9s
deployment-nginx-56d786cd98-xkjhq   0/1     ContainerCreating   0          3s
deployment-nginx-868855d887-2zh45   1/1     Running             0          3m22s
deployment-nginx-868855d887-hb6mv   1/1     Terminating         0          3m22s
deployment-nginx-868855d887-v8ndt   1/1     Terminating         0          3m16s

[root@server1 ~]# kubectl get pod
NAME                                READY   STATUS    RESTARTS   AGE
deployment-nginx-56d786cd98-78m7z   1/1     Running   0          26s
deployment-nginx-56d786cd98-8hnns   1/1     Running   0          31s
deployment-nginx-56d786cd98-lcvzw   1/1     Running   0          31s
deployment-nginx-56d786cd98-xkjhq   1/1     Running   0          25s

可以看出原来v1版本的rs被重新启用,再原来的rs下面新建4个pod。当然我门也可以使用kubectl delete rs --all命令删除不用的rs(注意:正在使用的rs不会删除):

[root@server1 ~]# kubectl delete rs --all
replicaset.apps "deployment-nginx-56d786cd98" deleted
replicaset.apps "deployment-nginx-868855d887" deleted
[root@server1 ~]# kubectl get pod
NAME                                READY   STATUS    RESTARTS   AGE
deployment-nginx-56d786cd98-6r589   1/1     Running   0          21s
deployment-nginx-56d786cd98-hvz24   1/1     Running   0          21s
deployment-nginx-56d786cd98-k62lj   1/1     Running   0          21s
deployment-nginx-56d786cd98-ss97z   1/1     Running   0          21s
[root@server1 ~]# kubectl get rs
NAME                          DESIRED   CURRENT   READY   AGE
deployment-nginx-56d786cd98   4         4         4       24s		#正在使用的RS不会删除

实验后删除:

[root@server1 ~]# kubectl delete -f deployment.yaml 
deployment.apps "deployment-nginx" deleted

DaemonSet举例

DaemonSet控制器保证每个节点上都运行一个pod:

[root@server1 ~]# vim daemonset.yaml 
[root@server1 ~]# cat daemonset.yaml 
apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: daemonset-example
  labels:
    app: zabbix-agent
spec:
  selector:
    matchLabels:
      name: zabbix-agent
  template:
    metadata:
      labels:
        name: zabbix-agent
    spec:
      containers:
      - name: zabbix-agent
        image: zabbix/zabbix-agent

创建pod:

[root@server1 ~]# kubectl apply -f daemonset.yaml 
daemonset.apps/daemonset-example created
[root@server1 ~]# kubectl get pod
NAME                      READY   STATUS    RESTARTS   AGE
daemonset-example-ctbgh   1/1     Running   0          6m5s
daemonset-example-mdqzk   1/1     Running   0          6m5s

[root@server1 ~]# kubectl get pod -o wide
NAME                      READY   STATUS    RESTARTS   AGE     IP            NODE      NOMINATED NODE   READINESS GATES
daemonset-example-ctbgh   1/1     Running   0          7m16s   10.244.2.32   server3   <none>           <none>
daemonset-example-mdqzk   1/1     Running   0          7m16s   10.244.1.25   server2   <none>           <none>

可以看出我们的每个节点(server3和server2)上都运行了一个pod,当我们删除一个pod时:

[root@server1 ~]# kubectl delete pod daemonset-example-ctbgh
pod "daemonset-example-ctbgh" deleted
[root@server1 ~]# kubectl get pod -o wide
NAME                      READY   STATUS              RESTARTS   AGE     IP            NODE      NOMINATED NODE   READINESS GATES
daemonset-example-998ck   0/1     ContainerCreating   0          8s      <none>        server3   <none>           <none>
daemonset-example-mdqzk   1/1     Running             0          8m22s   10.244.1.25   server2   <none>           <none>
[root@server1 ~]# kubectl get pod -o wide
NAME                      READY   STATUS    RESTARTS   AGE     IP            NODE      NOMINATED NODE   READINESS GATES
daemonset-example-998ck   1/1     Running   0          24s     10.244.2.33   server3   <none>           <none>
daemonset-example-mdqzk   1/1     Running   0          8m38s   10.244.1.25   server2   <none>           <none>

可以看出删除后DaemonSet控制器又会帮我们创建pod,以保证每个节点运行一个pod。

实验后删除:

[root@server1 ~]# kubectl delete -f daemonset.yaml 
daemonset.apps "daemonset-example" deleted

Job控制器举例

Job控制器只运行一次

[root@server1 ~]# vim job.yaml 
[root@server1 ~]# cat job.yaml 
apiVersion: batch/v1
kind: Job
metadata:
  name: pi
spec:
  template:
    spec:
      containers:
      - name: pi
        image: perl				#利用perl计算圆周率
        command: ["perl",  "-Mbignum=bpi", "-wle", "print bpi(2000)"]
      restartPolicy: Never
  backoffLimit: 4			#容器启动失败重启4次之后不再重启。

创建pod:

[root@server1 ~]# kubectl apply -f job.yaml 
job.batch/pi created
[root@server1 ~]# kubectl get pod
NAME       READY   STATUS              RESTARTS   AGE
pi-ktn6p   0/1     ContainerCreating   0          40s
[root@server1 ~]# kubectl get pod
NAME       READY   STATUS      RESTARTS   AGE
pi-ktn6p   0/1     Completed   0          2m12s

由于Job只运行一次,因此运行完后状态为Completed,我们可以查看日志获取结果
可以看出计算成功。

实验后删除:

[root@server1 ~]# kubectl delete -f job.yaml 
job.batch "pi" deleted

CronJob控制器举例

CronJob控制器用于定时执行任务:

[root@server1 ~]# vim cronjob.yaml 
[root@server1 ~]# cat cronjob.yaml 
apiVersion: batch/v1beta1
kind: CronJob
metadata:
  name: cronjob-example
spec:
  schedule: "* * * * *"
  jobTemplate:
    spec:
      template:
        spec:
          containers:
          - name: cronjob
            image: busybox
            args:
            - /bin/sh
            - -c
            - date; echo Hello from k8s cluster			#输出信息
          restartPolicy: OnFailure

其中schedule字段与crontab里面的写法相同,"* * * * *"表示每分钟执行任务。
创建pod:

[root@server1 ~]# kubectl apply -f cronjob.yaml 
cronjob.batch/cronjob-example created
root@server1 ~]# kubectl get pod
NAME                               READY   STATUS              RESTARTS   AGE
cronjob-example-1587387120-ngt6n   0/1     ContainerCreating   0          16s
[root@server1 ~]# kubectl get pod		#等待一分钟
NAME                               READY   STATUS              RESTARTS   AGE
cronjob-example-1587387120-ngt6n   0/1     Completed           0          60s
cronjob-example-1587387180-hbwzb   0/1     ContainerCreating   0          9s

可以看出每分钟运行一个pod,查看日志获取输出信息:

[root@server1 ~]# kubectl logs cronjob-example-1587387120-ngt6n
Mon Apr 20 12:52:30 UTC 2020
Hello from k8s cluster
[root@server1 ~]# kubectl logs cronjob-example-1587387180-hbwzb
Mon Apr 20 12:53:35 UTC 2020
Hello from k8s cluster

可以看出输出也是每分钟输出一次,也可以使用以下命令查看cronjob的信息:

[root@server1 ~]# kubectl get cronjobs
NAME              SCHEDULE    SUSPEND   ACTIVE   LAST SCHEDULE   AGE
cronjob-example   * * * * *   False     1        16s             3m10s

[root@server1 ~]# kubectl get job -w		#查看job信息并且持续输出
NAME                         COMPLETIONS   DURATION   AGE
cronjob-example-1587387120   1/1           20s        2m23s
cronjob-example-1587387180   1/1           35s        92s
cronjob-example-1587387240   1/1           20s        32s

实验后删除

[root@server1 ~]# kubectl delete -f cronjob.yaml 
cronjob.batch "cronjob-example" deleted