Kubernetes(5) Stateless Pod Controller -> ReplicaSet, DaemonSet, Deployment

最新推荐文章于 2022-12-15 22:27:52 发布

weixin_34419321

最新推荐文章于 2022-12-15 22:27:52 发布

阅读量232

点赞数

文章标签：网络数据库运维

原文链接：http://www.cnblogs.com/crazy-chinese/p/10240580.html

版权

1 Pod 控制器入门

ReplicaSet and Deployment (stateless) 可以管理无状态用户资源。Deployment是 ReplicaSet的 high level 版本实现，且封装了一些附加的常用功能。在实战中往往推荐使用Deployment而不是ReplicaSet。见 https://kubernetes.io/docs/concepts/workloads/controllers/replicaset/

DaemonSet (stateless) : 可以确保Pod能平均分布在每一个定义好的节点上启动，这种类型的控制器可以用作管理一些服务如： Beats 这样收集节点日志的服务

Jobs, CronJob: 一次性，或定期完成特定工作

StatefulSets, Operator -> 特殊用于 stateful 应用

2 ReplicaSet

ReplicaSet中有三个重要配置即：

replicas: 需要运行的副本数，可以设置为0，默认为1
selector: 按照标签j键值筛选匹配上的Pod 结果。标签的键值必须匹配才能控制到相应ReplicaSet，如果选择结果没有匹配，则返回空对象。详见: https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/#label-selectors .

template: 模块用于定义Pod，包括Pod的名字，Pod拥有的label以及Pod中运行的应用详见: https://kubernetes.io/docs/concepts/workloads/controllers/replicationcontroller#pod-template

一个例子用于展示如何创建 ReplicaSet，其中 spec.selector.matchLabels 决定哪个 Pod 会被选中如标签含有-> app: rs-demo-app 与 release: canary

spec.template.metadata.labels 定义了Pod 的标签.

使用如下模板生成2 副本的 ReplicaSet：

apiVersion: apps/v1
kind: ReplicaSet
metadata:
  name: rs-demo-app
  namespace: default
spec:
  replicas: 2
  selector:
    matchLabels:
      app: rs-demo-app
      release: canary
  template:
    metadata:
      name: rs-demo-app-pod
      labels:
        app: rs-demo-app
        release: canary
        enviroment: qa
    spec:
      containers:
        - name: rs-demo-app-container
          image: ikubernetes/myapp:v1
          ports:
          - name: http
            containerPort: 80
          livenessProbe:
            httpGet:
              port: http
              path: /index.html
            initialDelaySeconds: 1
            periodSeconds: 3

命令行演示:

[root@k8smaster controllers]# kubectl create -f replicaset-demo.yaml
replicaset.apps/rs-demo-app created
[root@k8smaster controllers]# kubectl get rs
NAME               DESIRED   CURRENT   READY   AGE
nginx-79976cbb47   3         3         3       4d
nginx-f4bd648c7    0         0         0       4d
rs-demo-app        2         2         2       12s
[root@k8smaster controllers]# kubectl get pods
NAME                     READY   STATUS    RESTARTS   AGE
nginx-79976cbb47-8dqnk   1/1     Running   1          4d
nginx-79976cbb47-p247g   1/1     Running   1          4d
nginx-79976cbb47-ppbqv   1/1     Running   1          4d
rs-demo-app-c2crz        1/1     Running   0          26s
rs-demo-app-xzddj        1/1     Running   0          26s


[root@k8smaster controllers]# kubectl describe rs-demo-app-c2crz
error: the server doesn't have a resource type "rs-demo-app-c2crz"


[root@k8smaster controllers]# kubectl describe po rs-demo-app-c2crz
Name:               rs-demo-app-c2crz
Namespace:          default
Priority:           0
PriorityClassName:  <none>
Node:               k8snode2/172.16.0.13
Start Time:         Mon, 07 Jan 2019 22:44:25 +0100
Labels:             app=rs-demo-app
                    enviroment=qa
                    release=canary
Annotations:        <none>
Status:             Running
IP:                 10.244.2.17
Controlled By:      ReplicaSet/rs-demo-app
Containers:
  rs-demo-app-container:
    Container ID:   docker://b82932a65bf6f02ccbb05ef6d39d2c27f42e90cc9a0e4a21924928133e29f9fa
    Image:          ikubernetes/myapp:v1
    Image ID:       docker-pullable://ikubernetes/myapp@sha256:9c3dc30b5219788b2b8a4b065f548b922a34479577befb54b03330999d30d513
    Port:           80/TCP
    Host Port:      0/TCP
    State:          Running
      Started:      Mon, 07 Jan 2019 22:44:26 +0100
    Ready:          True
    Restart Count:  0
    Liveness:       http-get http://:http/index.html delay=1s timeout=1s period=3s #success=1 #failure=3
    Environment:    <none>
    Mounts:
      /var/run/secrets/kubernetes.io/serviceaccount from default-token-rxs5t (ro)
Conditions:
  Type              Status
  Initialized       True 
  Ready             True 
  ContainersReady   True 
  PodScheduled      True 
Volumes:
  default-token-rxs5t:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  default-token-rxs5t
    Optional:    false
QoS Class:       BestEffort
Node-Selectors:  <none>
Tolerations:     node.kubernetes.io/not-ready:NoExecute for 300s
                 node.kubernetes.io/unreachable:NoExecute for 300s
Events:
  Type    Reason     Age   From               Message
  ----    ------     ----  ----               -------
  Normal  Scheduled  104s  default-scheduler  Successfully assigned default/rs-demo-app-c2crz to k8snode2
  Normal  Pulled     88s   kubelet, k8snode2  Container image "ikubernetes/myapp:v1" already present on machine
  Normal  Created    88s   kubelet, k8snode2  Created container
  Normal  Started    87s   kubelet, k8snode2  Started container

3 手动给 ReplicaSet 扩容

给ReplicaSet 扩容可以用 kubectl api 直接编辑。比如这里我们使用了 kubectl edit rs 然后将 spec.replicas 从2 上调至 3，这样会自动启动一个新的副本。

[root@k8smaster controllers]# kubectl edit rs rs-demo-app
replicaset.extensions/rs-demo-app edited # then change spec.replicas from 2 to 3

[root@k8smaster controllers]# kubectl get pod --show-labels
NAME                     READY   STATUS    RESTARTS   AGE   LABELS
nginx-79976cbb47-8dqnk   1/1     Running   1          4d    pod-template-hash=3553276603,run=nginx
nginx-79976cbb47-p247g   1/1     Running   1          4d    pod-template-hash=3553276603,run=nginx
nginx-79976cbb47-ppbqv   1/1     Running   1          4d    pod-template-hash=3553276603,run=nginx
rs-demo-app-c2crz        1/1     Running   0          14m   app=rs-demo-app,enviroment=qa,release=canary
rs-demo-app-lkh5j        1/1     Running   0          9s    app=rs-demo-app,enviroment=qa,release=canary
rs-demo-app-xzddj        1/1     Running   0          14m   app=rs-demo-app,enviroment=qa,release=canary

* 同理也可以通过修改 image 字段来更新应用的版本号，后面的文章会逐渐介绍这以功能

4 Deployments

Deployments 是加强版的 ReplicatSet 且拥有ReplicaSet的所有功能。也就是说，Deployment 可以用于执行一系列 ReplicaSet 组合操作如：Rollingupdate, 各种类型发布 (blue/green, canary etc.).

下面列举了 Deployment 的应用场景:

下面的例子展示了部署Deployment 且配置两个replica 副本:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: myapp-deploy
  namespace: default
spec:
  replicas: 2
  selector:
    matchLabels:
      app: myapp
      release: canary
  template:
    metadata:
      labels:
        app: myapp
        release: canary
    spec:
      containers:
        - name: myapp-container
          image: ikubernetes/myapp.v1
          ports:
            - name: http
              containerPort: 80

部署命令:

kubectl apply -f deployments-demo.yaml

检查部署情况:

[root@k8smaster controllers]# kubectl get deploy -o wide
NAME           DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS        IMAGES                 SELECTOR
myapp-deploy   2         2         2            0           5m    myapp-container   ikubernetes/myapp.v1   app=myapp,release=canary

[root@k8smaster controllers]# kubectl get rs -o wide
NAME                      DESIRED   CURRENT   READY   AGE   CONTAINERS        IMAGES                 SELECTOR
myapp-deploy-8546d69b76   2         2         0       5m    myapp-container   ikubernetes/myapp.v1   app=myapp,pod-template-hash=4102825632,release=canary

[root@k8smaster controllers]# kubectl get po -o wide
NAME                           READY   STATUS    RESTARTS   AGE   IP            NODE
myapp-deploy-f5f97bb54-2tlh6   1/1     Running   0          18s   10.244.2.23   k8snode2
myapp-deploy-f5f97bb54-65ffd   1/1     Running   0          18s   10.244.1.37   k8snode1

* 部署成功后发现，经过我们只写了部署脚本，但是同时自动生成了ReplicaSet对象，这也充分证明Deployment是ReplicaSet的升级版

* 名字: myapp-deploy-8546d69b76 来源于Deployment名字加 Pod 自动生成的 hash-code

5 Deployments 扩容

可以用修改模板然后靠 apply 命令完成自动扩容，比如修改模板中的 replicas 参数从 2 到 3 然后部署:

kubectl apply -f deployment-demo.yaml

与ReplicaSet不同用户可以重复使用这个命令来进行 Update操作，且不需要删除现有的 ReplicaSet，这个操作已经封装在 Deployment中了。

[root@k8smaster controllers]# kubectl describe deploy myapp-deploy
Name:                   myapp-deploy
Namespace:              default
CreationTimestamp:      Tue, 08 Jan 2019 06:02:20 +0100
Labels:                 <none>
Annotations:            deployment.kubernetes.io/revision: 1
                        kubectl.kubernetes.io/last-applied-configuration:
                          {"apiVersion":"apps/v1","kind":"Deployment","metadata":{"annotations":{},"name":"myapp-deploy","namespace":"default"},"spec":{"replicas":2...
Selector:               app=myapp,release=canary
Replicas:               2 desired | 2 updated | 2 total | 2 available | 0 unavailable
StrategyType:           RollingUpdate
MinReadySeconds:        0
RollingUpdateStrategy:  25% max unavailable, 25% max surge
Pod Template:
  Labels:  app=myapp
           release=canary
  Containers:
   myapp-container:
    Image:        ikubernetes/myapp:v1
    Port:         80/TCP
    Host Port:    0/TCP
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Conditions:
  Type           Status  Reason
  ----           ------  ------
  Available      True    MinimumReplicasAvailable
  Progressing    True    NewReplicaSetAvailable
OldReplicaSets:  <none>
NewReplicaSet:   myapp-deploy-f5f97bb54 (2/2 replicas created)
Events:
  Type    Reason             Age    From                   Message
  ----    ------             ----   ----                   -------
  Normal  ScalingReplicaSet  7m36s  deployment-controller  Scaled up replica set myapp-deploy-f5f97bb54 to 2

* if you change the template and deploy it again, kubernetes will remark the change information in Annotations -> like if you set replicas from 2 to 3...

* StrategyType by default is RollingUpdate and RollingUpdateStrategy: 25% max unavailable, 25% max surge

6 Deployments Update

现在我们修改模板的 replicas 从 2 到3， app version 从ikubernetes/myapp:v1 到 ikubernetes/myapp:v2 ，之后让这个修改生效，观察 Deployment的变动.

另开一个窗口用 -w 命令观察 Pod 被终止和重启的过程:

[root@k8smaster ~]# kubectl get pod -l app=myapp -w
NAME                           READY   STATUS    RESTARTS   AGE
myapp-deploy-f5f97bb54-2tlh6   1/1     Running   0          18m
myapp-deploy-f5f97bb54-65ffd   1/1     Running   0          18m

回之前的窗口执行部署命令：

[root@k8smaster controllers]# kubectl apply -f deployments-demo.yaml 
deployment.apps/myapp-deploy configured

我们可以看到新窗口的变化:

[root@k8smaster ~]# kubectl get pod -l app=myapp -w
NAME                           READY   STATUS    RESTARTS   AGE
myapp-deploy-f5f97bb54-2tlh6   1/1     Running   0          18m
myapp-deploy-f5f97bb54-65ffd   1/1     Running   0          18m
myapp-deploy-f5f97bb54-7lhmz   0/1   Pending   0     0s
myapp-deploy-f5f97bb54-7lhmz   0/1   Pending   0     0s
myapp-deploy-f5f97bb54-7lhmz   0/1   ContainerCreating   0     0s
myapp-deploy-f5f97bb54-7lhmz   1/1   Running   0     2s
myapp-deploy-5c574dbf-trf5k   0/1   Pending   0     0s
myapp-deploy-5c574dbf-trf5k   0/1   Pending   0     0s
myapp-deploy-5c574dbf-trf5k   0/1   ContainerCreating   0     0s
myapp-deploy-5c574dbf-trf5k   1/1   Running   0     5s
myapp-deploy-f5f97bb54-7lhmz   1/1   Terminating   0     7m12s
myapp-deploy-5c574dbf-p7jpw   0/1   Pending   0     0s
myapp-deploy-5c574dbf-p7jpw   0/1   Pending   0     0s
myapp-deploy-5c574dbf-p7jpw   0/1   ContainerCreating   0     0s
myapp-deploy-f5f97bb54-7lhmz   0/1   Terminating   0     7m14s
myapp-deploy-f5f97bb54-7lhmz   0/1   Terminating   0     7m15s
myapp-deploy-f5f97bb54-7lhmz   0/1   Terminating   0     7m15s
myapp-deploy-5c574dbf-p7jpw   1/1   Running   0     4s
myapp-deploy-f5f97bb54-2tlh6   1/1   Terminating   0     28m
myapp-deploy-5c574dbf-j25bz   0/1   Pending   0     0s
myapp-deploy-5c574dbf-j25bz   0/1   Pending   0     0s
myapp-deploy-5c574dbf-j25bz   0/1   ContainerCreating   0     0s
myapp-deploy-f5f97bb54-2tlh6   0/1   Terminating   0     28m
myapp-deploy-5c574dbf-j25bz   1/1   Running   0     1s
myapp-deploy-f5f97bb54-65ffd   1/1   Terminating   0     28m
myapp-deploy-f5f97bb54-65ffd   0/1   Terminating   0     28m
myapp-deploy-f5f97bb54-2tlh6   0/1   Terminating   0     28m
myapp-deploy-f5f97bb54-2tlh6   0/1   Terminating   0     28m
myapp-deploy-f5f97bb54-65ffd   0/1   Terminating   0     28m
myapp-deploy-f5f97bb54-65ffd   0/1   Terminating   0     28m

通过上面信息可以看到 Deployment 会自动创建一个新 Pod 然后干掉一个旧的

Update结束后会达到如下状态:

[root@k8smaster ~]# kubectl get rs -o wide
NAME                     DESIRED   CURRENT   READY   AGE   CONTAINERS        IMAGES                 SELECTOR
myapp-deploy-5c574dbf    3         3         3       50m   myapp-container   ikubernetes/myapp:v2   app=myapp,pod-template-hash=17130869,release=canary
myapp-deploy-f5f97bb54   0         0         0       1h    myapp-container   ikubernetes/myapp:v1   app=myapp,pod-template-hash=919536610,release=canary

[root@k8smaster ~]# kubectl get deploy -o wide
NAME           DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS        IMAGES                 SELECTOR
myapp-deploy   3         3         3            3           1h    myapp-container   ikubernetes/myapp:v2   app=myapp,release=canary

* ReplicaSet myapp-deploy-f5f97bb54 下面没有任何 Pods 新部署的ReplicaSet myapp-deploy-5c574dbf 升级到了新的版本

* 旧 ReplicaSet 依然会被显示，在特定情况下可以使用 Rollback 返回到上一个版本

用户也可以通过使用 kubectl rollout history deploy <deployment name> 检查版本信息:

[root@k8smaster ~]# kubectl rollout history deploy myapp-deploy
deployment.extensions/myapp-deploy 
REVISION  CHANGE-CAUSE
1         <none>
2         <none>

使用 kubectl rollout undo 可以做回滚操作，默认会返回上一版本:

[root@k8smaster ~]# kubectl get rs -o wide
NAME                     DESIRED   CURRENT   READY   AGE   CONTAINERS        IMAGES                 SELECTOR
myapp-deploy-5c574dbf    0         0         0       1h    myapp-container   ikubernetes/myapp:v2   app=myapp,pod-template-hash=17130869,release=canary
myapp-deploy-f5f97bb54   3         3         3       1h    myapp-container   ikubernetes/myapp:v1   app=myapp,pod-template-hash=919536610,release=canary

[root@k8smaster ~]# kubectl get deploy -o wide
NAME           DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS        IMAGES                 SELECTOR
myapp-deploy   3         3         3            3           1h    myapp-container   ikubernetes/myapp:v1   app=myapp,release=canary

现在我们回滚到 v2 使用命令 kubectl apply -f deployments-demo.yam...

7 Deployments Patch

也可以使用另一种方法升级或者扩容 Deploymets -> kubectl patch + json<content>

比如现在增加副本数从 3 到 5:

[root@k8smaster ~]# kubectl patch deployment myapp-deploy -p '{"spec":{"replicas":5}}'
deployment.extensions/myapp-deploy patched

[root@k8smaster ~]# kubectl get pods -o wide
NAME                          READY   STATUS    RESTARTS   AGE   IP            NODE
myapp-deploy-5c574dbf-5q6vm   1/1     Running   0          14m   10.244.1.42   k8snode1
myapp-deploy-5c574dbf-7t9qw   1/1     Running   0          8s    10.244.2.29   k8snode2
myapp-deploy-5c574dbf-p9mfb   1/1     Running   0          14m   10.244.1.41   k8snode1
myapp-deploy-5c574dbf-ssxkc   1/1     Running   0          8s    10.244.1.43   k8snode1
myapp-deploy-5c574dbf-tcbhz   1/1     Running   0          14m   10.244.2.28   k8snode2

8 用Deployment部署 Canary Release

1. 编写新的部署模板

apiVersion: apps/v1
kind: Deployment
metadata:
  name: myapp-deploy
  namespace: default
spec:
  replicas: 3
  selector:
    matchLabels:
      app: myapp
      release: canary
  template:
    metadata:
      labels:
        app: myapp
        release: canary
    spec:
      containers:
        - name: myapp-container
          image: ikubernetes/myapp:v2
          ports:
            - name: http
              containerPort: 80
  strategy:
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 0

2. 将镜像版本从 v2 变为 v3

3. pause Deployments:

# start rollout
kubectl apply -f deployment-demo.yaml

# pause after the first update been performed
kubectl rollout pause deployment myapp-deploy

# till here kubernetes will update one from Deployments-Pod

4. 恢复 Rollout Update

没有"pause" k8s将会按顺序删除一个Pod 然后添加一个新的 Pod 用于Update
但是现在有了 Pause 我们停止在当前状态下，该状态经过观察一共有 6 个Pod 生产环境中这个方法可以用于测试，比如在10个小时之后如果Update的版本工作顺利，则可以继续进行发布，使用以下命令：

kubectl rollout resume deployment myapp-deploy

这个命令会更新剩下的 Pods 到新的版本。最后可以使用后边的命令来检查当前的更新状态

kubectl get rs -o wide


9 Rollback
你可以使用 kubectl rollout undo 回滚到上一个版本

kubectl rollout undo deployment myapp-deploy

或者回滚至特定版本，你可以用以下命令来展示版本历史

kubectl rollout history deployment myapp-deploy

记住版本号，然后用这个版本号回滚至之前的状态

kubectl rollout undo deployment myapp-deploy --to-revision=1

10 DaemonSet

DeamonSet 保证所有或者被特定标记的节点上必定运行特定Pod 资源。当新节点加入集群时，Pod 也会相应添加进来。在节点被移除时，Pod 也会被相应地回收。在删除DaemonSet时，Pod 会随之被删除。

用下面模板创建 DaemonSet on Kubernetes Cluster

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: myapp-ds
  namespace: default
spec:
  selector:
    matchLabels:
      app: filebeat
      release: stable
  template:
    metadata:
      labels:
        app: filebeat
        release: stable
    spec:
      containers:
        - name: filebeat
          image: ikubernetes/filebeat:v5.6.5-alpine
          env:
            - name: REDIS_HOST
              value: redis.default.svc.cluster.local
            - name: REDIS_LOG_LEVEL
              value: info