了解ReplicationController

ReplicationController 是一种Kubernetes 资源，简称RC，可确保它的pod 始终保持运行状态。如果pod 因任何原因消失（例如节点从集群中消失或由于该pod 从节点中逐出），则ReplicationController 会注意到缺少了pod 并创建替代pod 。

06.ReplicationController - 图1

当一个节点突然下线且带有两个pod 时会发生什么。pod A 是被直接创建的，因此是非托管的pod ，而pod B 由ReplicationController 管理。节点异常退出后，ReplicationController 会创建一个新的pod ( pod B2 ）来替换缺少的pod B ，而pod A完全丢失一一没有东西负责重建它。

ReplicationController 会持续监控正在运行的pod 列表，并保证相应“类型”的pod 的数目与期望相符。如正在运行的pod 太少，它会根据pod 模板创建新的副本。如正在运行的pod 太多，它将删除多余的副本。

06.ReplicationController - 图2

ReplicationController 操作

一个ReplicationController 有三个主要部分

  • label selector （标签选择器） ，用于确定ReplicationController 作用域中有哪些pod
  • replica count （副本个数），指定应运行的pod 数量
  • pod template ( pod 模板），用于创建新的pod 副本
  1. [root@master01 ~]# kubectl explain ReplicationController
  2. KIND:     ReplicationController
  3. VERSION:  v1
  5. DESCRIPTION:
  6.      ReplicationController represents the configuration of a replication
  7.      controller.
  9. FIELDS:
  10.    apiVersion    <string>
  11.      APIVersion defines the versioned schema of this representation of an
  12.      object. Servers should convert recognized schemas to the latest internal
  13.      value, and may reject unrecognized values. More info:
  14.      https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources
  16.    kind    <string>
  17.      Kind is a string value representing the REST resource this object
  18.      represents. Servers may infer this from the endpoint the client submits
  19.      requests to. Cannot be updated. In CamelCase. More info:
  20.      https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds
  22.    metadata    <Object>
  23.      If the Labels of a ReplicationController are empty, they are defaulted to
  24.      be the same as the Pod(s) that the replication controller manages. Standard
  25.      object's metadata. More info:
  26.      https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#metadata
  28.    spec    <Object>
  29.      Spec defines the specification of the desired behavior of the replication
  30.      controller. More info:
  31.      https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status
  33.    status    <Object>
  34.      Status is the most recently observed status of the replication controller.
  35.      This data may be out of date by some window of time. Populated by the
  36.      system. Read-only. More info:
  37.      https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status
  1. [root@master01 ~]# kubectl explain ReplicationController.spec
  2. KIND:     ReplicationController
  3. VERSION:  v1
  5. RESOURCE: spec <Object>
  7. DESCRIPTION:
  8.      Spec defines the specification of the desired behavior of the replication
  9.      controller. More info:
  10.      https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status
  12.      ReplicationControllerSpec is the specification of a replication controller.
  14. FIELDS:
  15.    minReadySeconds    <integer>
  16.      Minimum number of seconds for which a newly created pod should be ready
  17.      without any of its container crashing, for it to be considered available.
  18.      Defaults to 0 (pod will be considered available as soon as it is ready)
  20.    replicas    <integer>
  21.      Replicas is the number of desired replicas. This is a pointer to
  22.      distinguish between explicit zero and unspecified. Defaults to 1. More
  23.      info:
  24.      https://kubernetes.io/docs/concepts/workloads/controllers/replicationcontroller#what-is-a-replicationcontroller
  26.    selector    <map[string]string>
  27.      Selector is a label query over pods that should match the Replicas count.
  28.      If Selector is empty, it is defaulted to the labels present on the Pod
  29.      template. Label keys and values that must match in order to be controlled
  30.      by this replication controller, if empty defaulted to labels on Pod
  31.      template. More info:
  32.      https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/#label-selectors
  34.    template    <Object>
  35.      Template is the object that describes the pod that will be created if
  36.      insufficient replicas are detected. This takes precedence over a
  37.      TemplateRef. More info:
  38.      https://kubernetes.io/docs/concepts/workloads/controllers/replicationcontroller#pod-template

创建RC

  1. cat > myapp-rc.yaml <<EOF
  2. apiVersion: v1
  3. kind: ReplicationController
  4. metadata:
  5.   name: myapp-rc
  6.   labels:
  7.     app: myapp
  8.     type: rc
  9. spec:
  10.   replicas: 3
  11.   selector:
  12.       app: myapp
  13.       type: rc
  14.   template:
  15.     metadata:
  16.       name: myapp-pod
  17.       labels:
  18.         app: myapp
  19.         type: rc
  20.     spec:
  21.       containers:
  22.       - name: myapp
  23.         image: ikubernetes/myapp:v1
  24.         ports:
  25.         - name: http
  26.           containerPort: 80
  27. EOF

上面的YAML文件：

  • apiVersion: v1
  • kindRe: plicationController
  • spec.replicas: 指定Pod副本数量，默认为1
  • spec.selector: RC通过该属性来筛选要控制的Pod
  • spec.template: 这里就是我们之前的Pod的定义的模块，但是不需要apiVersion和kind了
  • spec.template.metadata.labels: 注意这里的Pod的labels要和spec.selector相同，这样RC就可以来控制当前这个Pod了。
  1. kubectl apply -f myapp-rc.yaml
  1. [root@master01 ~]# kubectl get pod -o wide
  2. NAME             READY   STATUS    RESTARTS   AGE   IP               NODE     NOMINATED NODE   READINESS GATES
  3. myapp-rc-2nh88   1/1     Running   0          43s   10.244.186.204   node03   <none>           <none>
  4. myapp-rc-dl8px   1/1     Running   0          43s   10.244.196.139   node01   <none>           <none>
  5. myapp-rc-vhbm7   1/1     Running   0          43s   10.244.140.71    node02   <none>           <none>

查看rc

[root@master01 ~]# kubectl get rc
NAME       DESIRED   CURRENT   READY   AGE
myapp-rc   3         3         3       3m59s

查看具体信息

[root@master01 ~]# kubectl describe rc myapp-rc 
Name:         myapp-rc
Namespace:    default
Selector:     app=myapp,type=rc
Labels:       app=myapp
              type=rc
Annotations:  kubectl.kubernetes.io/last-applied-configuration:
                {"apiVersion":"v1","kind":"ReplicationController","metadata":{"annotations":{},"labels":{"app":"myapp","type":"rc"},"name":"myapp-rc","nam...
Replicas:     3 current / 3 desired
Pods Status:  3 Running / 0 Waiting / 0 Succeeded / 0 Failed
Pod Template:
  Labels:  app=myapp
           type=rc
  Containers:
   myapp:
    Image:        ikubernetes/myapp:v1
    Port:         80/TCP
    Host Port:    0/TCP
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Events:
  Type    Reason            Age    From                    Message
  ----    ------            ----   ----                    -------
  Normal  SuccessfulCreate  4m47s  replication-controller  Created pod: myapp-rc-dl8px
  Normal  SuccessfulCreate  4m47s  replication-controller  Created pod: myapp-rc-2nh88
  Normal  SuccessfulCreate  4m46s  replication-controller  Created pod: myapp-rc-vhbm7

修改RC

vim myapp-rc.yaml
....
replicas: 5

kubectl apply -f myapp-rc.yaml

[root@master01 ~]# kubectl get rc
NAME       DESIRED   CURRENT   READY   AGE
myapp-rc   5         5         5       7m41s
[root@master01 ~]# kubectl get pod -o wide
NAME             READY   STATUS    RESTARTS   AGE     IP               NODE     NOMINATED NODE   READINESS GATES
myapp-rc-2nh88   1/1     Running   0          7m50s   10.244.186.204   node03   <none>           <none>
myapp-rc-dl8px   1/1     Running   0          7m50s   10.244.196.139   node01   <none>           <none>
myapp-rc-f4qzt   1/1     Running   0          21s     10.244.186.205   node03   <none>           <none>
myapp-rc-rqs74   1/1     Running   0          21s     10.244.196.140   node01   <none>           <none>
myapp-rc-vhbm7   1/1     Running   0          7m50s   10.244.140.71    node02   <none>           <none>

删除rc中的一个pod

[root@master01 ~]# kubectl delete pod myapp-rc-2nh88
pod "myapp-rc-2nh88" deleted

[root@master01 ~]# kubectl get pod -w
NAME             READY   STATUS    RESTARTS   AGE
myapp-rc-2nh88   1/1     Running   0          10m
myapp-rc-dl8px   1/1     Running   0          10m
myapp-rc-f4qzt   1/1     Running   0          3m2s
myapp-rc-rqs74   1/1     Running   0          3m2s
myapp-rc-vhbm7   1/1     Running   0          10m
myapp-rc-2nh88   1/1     Terminating   0          10m
myapp-rc-29rwv   0/1     Pending       0          0s
myapp-rc-29rwv   0/1     Pending       0          0s
myapp-rc-29rwv   0/1     ContainerCreating   0          0s
myapp-rc-29rwv   0/1     ContainerCreating   0          1s
myapp-rc-29rwv   1/1     Running             0          2s

06.ReplicationController - 图3

模拟节点故障

将node03 网卡下线

[root@node03 ~]# ifconfig eth0  down

观察pod是否迁移

[root@master01 ~]# kubectl get node
NAME       STATUS     ROLES    AGE   VERSION
master01   Ready      master   44h   v1.16.4
master02   Ready      master   44h   v1.16.4
master03   Ready      master   44h   v1.16.4
node01     Ready      node     44h   v1.16.4
node02     Ready      node     44h   v1.16.4
node03     NotReady   node     44h   v1.16.4

[root@master01 ~]# kubectl get pod -w -o wide
NAME             READY   STATUS    RESTARTS   AGE     IP               NODE     NOMINATED NODE   READINESS GATES
myapp-rc-29rwv   1/1     Running   0          8m19s   10.244.186.206   node03   <none>           <none>
myapp-rc-dl8px   1/1     Running   0          19m     10.244.196.139   node01   <none>           <none>
myapp-rc-f4qzt   1/1     Running   0          11m     10.244.186.205   node03   <none>           <none>
myapp-rc-rqs74   1/1     Running   0          11m     10.244.196.140   node01   <none>           <none>
myapp-rc-vhbm7   1/1     Running   0          19m     10.244.140.71    node02   <none>           <none>
myapp-rc-f4qzt   1/1     Terminating   0          14m     10.244.186.205   node03   <none>           <none>
myapp-rc-29rwv   1/1     Terminating   0          11m     10.244.186.206   node03   <none>           <none>
myapp-rc-tsgjs   0/1     Pending       0          0s      <none>           <none>   <none>           <none>
myapp-rc-tsgjs   0/1     Pending       0          0s      <none>           node02   <none>           <none>
myapp-rc-nz9sj   0/1     Pending       0          0s      <none>           <none>   <none>           <none>
myapp-rc-nz9sj   0/1     Pending       0          0s      <none>           node01   <none>           <none>
myapp-rc-tsgjs   0/1     ContainerCreating   0          0s      <none>           node02   <none>           <none>
myapp-rc-nz9sj   0/1     ContainerCreating   0          0s      <none>           node01   <none>           <none>
myapp-rc-nz9sj   0/1     ContainerCreating   0          1s      <none>           node01   <none>           <none>
myapp-rc-tsgjs   0/1     ContainerCreating   0          1s      <none>           node02   <none>           <none>
myapp-rc-tsgjs   1/1     Running             0          2s      10.244.140.72    node02   <none>           <none>
myapp-rc-nz9sj   1/1     Running             0          2s      10.244.196.141   node01   <none>           <none>

[root@master01 ~]# kubectl get pods -o wide
NAME             READY   STATUS    RESTARTS   AGE   IP               NODE     NOMINATED NODE   READINESS GATES
myapp-rc-dl8px   1/1     Running   0          23m   10.244.196.139   node01   <none>           <none>
myapp-rc-nz9sj   1/1     Running   0          95s   10.244.196.141   node01   <none>           <none>
myapp-rc-rqs74   1/1     Running   0          16m   10.244.196.140   node01   <none>           <none>
myapp-rc-tsgjs   1/1     Running   0          95s   10.244.140.72    node02   <none>           <none>
myapp-rc-vhbm7   1/1     Running   0          23m   10.244.140.71    node02   <none>           <none>

将pod 移入或移出ReplicationController 的作用域

由ReplicationController 创建的pod 并不是绑定到ReplicationController 。在任何时刻， ReplicationController 管理与标签选择器匹配的pod 。通过更改pod 的标签， 可以将它从ReplicationController 的作用域中添加或删除。它甚至可以从一个ReplicationController 移动到另一个。

如果你更改了一个pod 的标签，使它不再与ReplicationController的标签选择器相匹配， 那么该pod 就变得和其他手动创建的pod 一样了。它不再被任何东西管理。如果运行该节点的pod 异常终止， 它显然不会被重新调度。但请记住，当你更改pod 的标签时， ReplicationController 发现一个pod 丢失了， 并启动一个新的pod替换它。

添加标签

[root@master01 ~]# kubectl get pod --show-labels 
NAME             READY   STATUS    RESTARTS   AGE     LABELS
myapp-rc-dl8px   1/1     Running   0          30m     app=myapp,type=rc
myapp-rc-nz9sj   1/1     Running   0          8m39s   app=myapp,type=rc
myapp-rc-rqs74   1/1     Running   0          23m     app=myapp,type=rc
myapp-rc-tsgjs   1/1     Running   0          8m39s   app=myapp,type=rc
myapp-rc-vhbm7   1/1     Running   0          30m     app=myapp,type=rc

[root@master01 ~]# kubectl get pod --show-labels 
NAME             READY   STATUS    RESTARTS   AGE     LABELS
myapp-rc-dl8px   1/1     Running   0          30m     app=myapp,type=rc
myapp-rc-nz9sj   1/1     Running   0          8m39s   app=myapp,type=rc
myapp-rc-rqs74   1/1     Running   0          23m     app=myapp,type=rc
myapp-rc-tsgjs   1/1     Running   0          8m39s   app=myapp,type=rc
myapp-rc-vhbm7   1/1     Running   0          30m     app=myapp,type=rc
[root@master01 ~]# kubectl label pod myapp-rc-dl8px release=beta
pod/myapp-rc-dl8px labeled
[root@master01 ~]# kubectl get pod --show-labels 
NAME             READY   STATUS    RESTARTS   AGE     LABELS
myapp-rc-dl8px   1/1     Running   0          31m     app=myapp,release=beta,type=rc
myapp-rc-nz9sj   1/1     Running   0          9m33s   app=myapp,type=rc
myapp-rc-rqs74   1/1     Running   0          24m     app=myapp,type=rc
myapp-rc-tsgjs   1/1     Running   0          9m33s   app=myapp,type=rc
myapp-rc-vhbm7   1/1     Running   0          31m     app=myapp,type=rc

给其中一个pod 添加了标签，再次列出所有pod 会显示和以前一样的三个pod 。因为从ReplicationController 角度而言， 没发生任何更改

删除标签

现在，删除一个标签。这将使该pod不再与ReplicationController 的标签选择器相匹配，只剩下4个匹配的pod 。因此， ReplicationController 会启动一个新的pod ， 将数目恢复为5：

[root@master01 ~]# kubectl label pod myapp-rc-nz9sj type-
pod/myapp-rc-nz9sj labeled

[root@master01 ~]# kubectl get pod --show-labels 
NAME             READY   STATUS    RESTARTS   AGE   LABELS
myapp-rc-dl8px   1/1     Running   0          33m   app=myapp,release=beta,type=rc
myapp-rc-nz9sj   1/1     Running   0          11m   app=myapp
myapp-rc-rqs74   1/1     Running   0          26m   app=myapp,type=rc
myapp-rc-rtm8c   1/1     Running   0          2s    app=myapp,type=rc
myapp-rc-tsgjs   1/1     Running   0          11m   app=myapp,type=rc
myapp-rc-vhbm7   1/1     Running   0          33m   app=myapp,type=rc

修改标签

现在，修改一个标签。这将使该pod不再与ReplicationController 的标签选择器相匹配，只剩下4个匹配的pod 。因此， ReplicationController 会启动一个新的pod ， 将数目恢复为5：

[root@master01 ~]# kubectl label pod myapp-rc-tsgjs  app=myapp01 --overwrite 
pod/myapp-rc-tsgjs labeled

[root@master01 ~]# kubectl get pod --show-labels -w
NAME             READY   STATUS    RESTARTS   AGE     LABELS
myapp-rc-dl8px   1/1     Running   0          38m     app=myapp,release=beta,type=rc
myapp-rc-nz9sj   1/1     Running   0          16m     app=myapp
myapp-rc-rqs74   1/1     Running   0          30m     app=myapp,type=rc
myapp-rc-rtm8c   1/1     Running   0          4m39s   app=myapp,type=rc
myapp-rc-tsgjs   1/1     Running   0          16m     app=myapp,type=rc
myapp-rc-vhbm7   1/1     Running   0          38m     app=myapp,type=rc
myapp-rc-tsgjs   1/1     Running   0          16m     app=myapp01,type=rc
myapp-rc-tsgjs   1/1     Running   0          16m     app=myapp01,type=rc
myapp-rc-27gd6   0/1     Pending   0          0s      app=myapp,type=rc
myapp-rc-27gd6   0/1     Pending   0          0s      app=myapp,type=rc
myapp-rc-27gd6   0/1     ContainerCreating   0          0s      app=myapp,type=rc
myapp-rc-27gd6   0/1     ContainerCreating   0          1s      app=myapp,type=rc
myapp-rc-27gd6   1/1     Running             0          3s      app=myapp,type=rc

[root@master01 ~]# kubectl get pod --show-labels
NAME             READY   STATUS    RESTARTS   AGE     LABELS
myapp-rc-27gd6   1/1     Running   0          66s     app=myapp,type=rc
myapp-rc-dl8px   1/1     Running   0          40m     app=myapp,release=beta,type=rc
myapp-rc-nz9sj   1/1     Running   0          17m     app=myapp
myapp-rc-rqs74   1/1     Running   0          32m     app=myapp,type=rc
myapp-rc-rtm8c   1/1     Running   0          6m26s   app=myapp,type=rc
myapp-rc-tsgjs   1/1     Running   0          17m     app=myapp01,type=rc
myapp-rc-vhbm7   1/1     Running   0          40m     app=myapp,type=rc

06.ReplicationController - 图4

扩容RC

[root@master01 ~]# kubectl scale rc myapp-rc --replicas=10
replicationcontroller/myapp-rc scaled

[root@master01 ~]# kubectl get pod --show-labels
NAME             READY   STATUS    RESTARTS   AGE     LABELS
myapp-rc-27gd6   1/1     Running   0          6m25s   app=myapp,type=rc
myapp-rc-5p7ws   1/1     Running   0          15s     app=myapp,type=rc
myapp-rc-66ggv   1/1     Running   0          15s     app=myapp,type=rc
myapp-rc-dl8px   1/1     Running   0          45m     app=myapp,release=beta,type=rc
myapp-rc-kdw6p   1/1     Running   0          15s     app=myapp,type=rc
myapp-rc-nz9sj   1/1     Running   0          23m     app=myapp
myapp-rc-pqqk4   1/1     Running   0          15s     app=myapp,type=rc
myapp-rc-rqs74   1/1     Running   0          38m     app=myapp,type=rc
myapp-rc-rtm8c   1/1     Running   0          11m     app=myapp,type=rc
myapp-rc-tsgjs   1/1     Running   0          23m     app=myapp01,type=rc
myapp-rc-vhbm7   1/1     Running   0          45m     app=myapp,type=rc
myapp-rc-w655s   1/1     Running   0          15s     app=myapp,type=rc

滚动更新

rolling-update官方已经不推荐了，因为RC现在已被deployment取代了，现在都是直接用deployment

[root@master01 ~]# kubectl rolling-update myapp-rc --image=ikubernetes/myapp:v2
Command "rolling-update" is deprecated, use "rollout" instead
Found existing update in progress (myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3), resuming.
Continuing update with existing controller myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3.
Scaling up myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 from 1 to 10, scaling down myapp-rc from 10 to 0 (keep 10 pods available, don't exceed 11 pods)
Scaling myapp-rc down to 9
Scaling myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 up to 2
Scaling myapp-rc down to 8
Scaling myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 up to 3
Scaling myapp-rc down to 7
Scaling myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 up to 4
Scaling myapp-rc down to 6
Scaling myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 up to 5
Scaling myapp-rc down to 5
Scaling myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 up to 6
Scaling myapp-rc down to 4
Scaling myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 up to 7
Scaling myapp-rc down to 3
Scaling myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 up to 8
Scaling myapp-rc down to 2
Scaling myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 up to 9
Scaling myapp-rc down to 1
Scaling myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 up to 10
Scaling myapp-rc down to 0
Update succeeded. Deleting old controller: myapp-rc
Renaming myapp-rc-9545d61ba1b2ce3cbe6eef6a8e8abef3 to myapp-rc
replicationcontroller/myapp-rc rolling updated

删除RC

[root@master01 ~]# kubectl delete rc myapp-rc 
replicationcontroller "myapp-rc" deleted