Pod和Deployment的使用(k8s实践-3)

使用Pod

Pod.yaml

apiVersion: v1
kind: Pod
metadata:
  name: hello-pod
  labels:
    zone: prod
    version: v1
spec:
  containers:
  - name: hello-ctr
    image: nigelpoulton/k8sbook:latest
    ports:
    - containerPort: 8080

创建Pod

root@lzl:/home/lzl# kubectl apply -f ./WorkSpace/k8s/pod/Pod.yml 
pod/hello-pod created
root@lzl:/home/lzl# kubectl get Pods
NAME        READY   STATUS              RESTARTS   AGE
hello-pod   0/1     ContainerCreating   0          12s

# 要等一会，因为有拉取镜像的时间
root@lzl:/home/lzl# kubectl get Pods
NAME        READY   STATUS    RESTARTS   AGE
hello-pod   1/1     Running   0          110s

获取Pod信息

获取Pod详细信息：

kubectl get Pods hello-pod -o yaml

root@lzl:/home/lzl# kubectl get Pods hello-pod -o yaml
apiVersion: v1
kind: Pod
metadata:
  annotations:
    kubectl.kubernetes.io/last-applied-configuration: |
      {"apiVersion":"v1","kind":"Pod","metadata":{"annotations":{},"labels":{"version":"v1","zone":"prod"},"name":"hello-pod","namespace":"default"},"spec":{"containers":[{"image":"nigelpoulton/k8sbook:latest","name":"hello-ctr","ports":[{"containerPort":8080}]}]}}
  creationTimestamp: "2022-01-19T10:26:35Z"
  labels:
    version: v1
    zone: prod
  name: hello-pod
  namespace: default
  resourceVersion: "29553"
  uid: 1f41e62a-a951-4593-83f7-f3e7398b9d24
spec:	# 这部分是Pod的期望状态
  containers:
  - image: nigelpoulton/k8sbook:latest
    imagePullPolicy: Always
    name: hello-ctr
    ports:
    - containerPort: 8080
      protocol: TCP
    resources: {}
    terminationMessagePath: /dev/termination-log
    terminationMessagePolicy: File
    volumeMounts:
    - mountPath: /var/run/secrets/kubernetes.io/serviceaccount
      name: kube-api-access-5kwmz
      readOnly: true
  dnsPolicy: ClusterFirst
  enableServiceLinks: true
  nodeName: lzl-c
  preemptionPolicy: PreemptLowerPriority
  priority: 0
  restartPolicy: Always
  schedulerName: default-scheduler
  securityContext: {}
  serviceAccount: default
  serviceAccountName: default
  terminationGracePeriodSeconds: 30
  tolerations:
  - effect: NoExecute
    key: node.kubernetes.io/not-ready
    operator: Exists
    tolerationSeconds: 300
  - effect: NoExecute
    key: node.kubernetes.io/unreachable
    operator: Exists
    tolerationSeconds: 300
  volumes:
  - name: kube-api-access-5kwmz
    projected:
      defaultMode: 420
      sources:
      - serviceAccountToken:
          expirationSeconds: 3607
          path: token
      - configMap:
          items:
          - key: ca.crt
            path: ca.crt
          name: kube-root-ca.crt
      - downwardAPI:
          items:
          - fieldRef:
              apiVersion: v1
              fieldPath: metadata.namespace
            path: namespace
status:	# 这部分是Pod的当前状态
  conditions:
  - lastProbeTime: null
    lastTransitionTime: "2022-01-19T10:26:35Z"
    status: "True"
    type: Initialized
  - lastProbeTime: null
    lastTransitionTime: "2022-01-19T10:26:58Z"
    status: "True"
    type: Ready
  - lastProbeTime: null
    lastTransitionTime: "2022-01-19T10:26:58Z"
    status: "True"
    type: ContainersReady
  - lastProbeTime: null
    lastTransitionTime: "2022-01-19T10:26:35Z"
    status: "True"
    type: PodScheduled
  containerStatuses:
  - containerID: docker://a7cc9815fc32e3f99f57553c6c28aa264576dbc3076784fee485a4306a1f4fcb
    image: nigelpoulton/k8sbook:latest
    imageID: docker-pullable://nigelpoulton/k8sbook@sha256:a983a96a85151320cd6ad0cd9fda3b725a743ed642e58b0597285c6bcb46c90f
    lastState: {}
    name: hello-ctr
    ready: true
    restartCount: 0
    started: true
    state:
      running:
        startedAt: "2022-01-19T10:26:58Z"
  hostIP: 192.168.230.13
  phase: Running
  podIP: 10.5.2.2
  podIPs:
  - ip: 10.5.2.2
  qosClass: BestEffort
  startTime: "2022-01-19T10:26:35Z"

kubectl describe Pods hello-pod

root@lzl:/home/lzl# kubectl describe Pods hello-pod
Name:         hello-pod
Namespace:    default
Priority:     0
Node:         lzl-c/192.168.230.13
Start Time:   Wed, 19 Jan 2022 18:26:35 +0800
Labels:       version=v1
              zone=prod
Annotations:  <none>
Status:       Running
IP:           10.5.2.2
IPs:
  IP:  10.5.2.2
Containers:
  hello-ctr:
    Container ID:   docker://a7cc9815fc32e3f99f57553c6c28aa264576dbc3076784fee485a4306a1f4fcb
    Image:          nigelpoulton/k8sbook:latest
    Image ID:       docker-pullable://nigelpoulton/k8sbook@sha256:a983a96a85151320cd6ad0cd9fda3b725a743ed642e58b0597285c6bcb46c90f
    Port:           8080/TCP
    Host Port:      0/TCP
    State:          Running
      Started:      Wed, 19 Jan 2022 18:26:58 +0800
    Ready:          True
    Restart Count:  0
    Environment:    <none>
    Mounts:
      /var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-5kwmz (ro)
Conditions:
  Type              Status
  Initialized       True 
  Ready             True 
  ContainersReady   True 
  PodScheduled      True 
Volumes:
  kube-api-access-5kwmz:
    Type:                    Projected (a volume that contains injected data from multiple sources)
    TokenExpirationSeconds:  3607
    ConfigMapName:           kube-root-ca.crt
    ConfigMapOptional:       <nil>
    DownwardAPI:             true
QoS Class:                   BestEffort
Node-Selectors:              <none>
Tolerations:                 node.kubernetes.io/not-ready:NoExecute op=Exists for 300s
                             node.kubernetes.io/unreachable:NoExecute op=Exists for 300s
Events:	# Pod生命周期中的一些重要事件
  Type    Reason     Age    From               Message
  ----    ------     ----   ----               -------
  Normal  Scheduled  6m56s  default-scheduler  Successfully assigned default/hello-pod to lzl-c
  Normal  Pulling    6m54s  kubelet            Pulling image "nigelpoulton/k8sbook:latest"
  Normal  Pulled     6m33s  kubelet            Successfully pulled image "nigelpoulton/k8sbook:latest" in 21.212884013s
  Normal  Created    6m33s  kubelet            Created container hello-ctr
  Normal  Started    6m33s  kubelet            Started container hello-ctr

进入Pod中的容器内部

还可以进入容器执行命令，来查看Pod的信息：kubectl exec hello-pod -- ps aux

root@lzl:/home/lzl# kubectl exec hello-pod -- ps aux
PID   USER     TIME  COMMAND
    1 root      0:00 node ./app.js
   13 root      0:00 ps aux

登录到运行容器内部：kubectl exec -it hello-pod -- sh

root@lzl:/home/lzl# kubectl exec -it hello-pod -- sh
/src # ls
Dockerfile         app.js             package-lock.json  views
README.md          node_modules       package.json

/src # exit
root@lzl:/home/lzl#

当Pod中有多个容器时，需要使用--container参数指定想要创建交互式会话的容器，不指定的话默认是第一个容器。

使用exit退出当前容器。

删除Pod

kubectl delete -f pod.yaml

root@lzl:/home/lzl# kubectl delete -f ./WorkSpace/k8s/pod/Pod.yml 
pod "hello-pod" deleted
root@lzl:/home/lzl# kubectl get Pods
No resources found in default namespace.

使用Deployment部署Pod

Deployment为Kubernetes带来自愈、自动扩缩容、滚动升级以及基于版本的回滚的能力。实际上在底层Deployment是使用的ReplicaSet来完成的，可以理解为，Deployment管理ReplicaSet，而ReplicaSet管理Pod。

部署Deployment

本次部署的文件deploy.yaml如下：

apiVersion: apps/v1
kind: Deployment
metadata:
  name: hello-deploy
spec:                       # spec下的内容都与Pod有关
  replicas: 10
  selector:                 # 表示该Deployment所管理的Pod必须要有的标签
    matchLabels:
      app: hello-world
  minReadySeconds: 10
  strategy:                 # 表示更新策略
    type: RollingUpdate
    rollingUpdate:
      maxUnavailable: 1
      maxSurge: 1
  template:                 # 所管理的Pod的模板，这里只有一个容器
    metadata:
      labels:
        app: hello-world
    spec:
      containers:
      - name: hello-pod
        image: nigelpoulton/k8sbook:latest
        ports:
        - containerPort: 8080

进行部署：

root@lzl:/home/lzl# kubectl apply -f ./WorkSpace/k8s/deployment/deploy.yaml 
deployment.apps/hello-deploy created

查看Deployment

使用kubectl get

root@lzl:/home/lzl# kubectl get deploy hello-deploy
NAME           READY   UP-TO-DATE   AVAILABLE   AGE
hello-deploy   10/10   10           10          2m29s

使用kubectl describe

root@lzl:/home/lzl# kubectl describe deploy hello-deploy
Name:                   hello-deploy
Namespace:              default
CreationTimestamp:      Wed, 19 Jan 2022 19:18:16 +0800
Labels:                 <none>
Annotations:            deployment.kubernetes.io/revision: 1
Selector:               app=hello-world
Replicas:               10 desired | 10 updated | 10 total | 10 available | 0 unavailable
StrategyType:           RollingUpdate
MinReadySeconds:        10
RollingUpdateStrategy:  1 max unavailable, 1 max surge
Pod Template:
  Labels:  app=hello-world
  Containers:
   hello-pod:
    Image:        nigelpoulton/k8sbook:latest
    Port:         8080/TCP
    Host Port:    0/TCP
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Conditions:
  Type           Status  Reason
  ----           ------  ------
  Available      True    MinimumReplicasAvailable
  Progressing    True    NewReplicaSetAvailable
OldReplicaSets:  <none>
NewReplicaSet:   hello-deploy-65cbc9474c (10/10 replicas created)
Events:
  Type    Reason             Age    From                   Message
  ----    ------             ----   ----                   -------
  Normal  ScalingReplicaSet  4m17s  deployment-controller  Scaled up replica set hello-deploy-65cbc9474c to 10

可以使用kubectl get rs来查看ReplicaSet的状态，以及kubectl describe rs来查看其详细信息

root@lzl:/home/lzl# kubectl get rs
NAME                      DESIRED   CURRENT   READY   AGE
hello-deploy-65cbc9474c   10        10        10      5m55s

root@lzl:/home/lzl# kubectl describe rs hello-deploy-65cbc9474c
Name:           hello-deploy-65cbc9474c
Namespace:      default
Selector:       app=hello-world,pod-template-hash=65cbc9474c
Labels:         app=hello-world
                pod-template-hash=65cbc9474c
Annotations:    deployment.kubernetes.io/desired-replicas: 10
                deployment.kubernetes.io/max-replicas: 11
                deployment.kubernetes.io/revision: 1
Controlled By:  Deployment/hello-deploy
Replicas:       10 current / 10 desired
Pods Status:    10 Running / 0 Waiting / 0 Succeeded / 0 Failed
Pod Template:
  Labels:  app=hello-world
           pod-template-hash=65cbc9474c
  Containers:
   hello-pod:
    Image:        nigelpoulton/k8sbook:latest
    Port:         8080/TCP
    Host Port:    0/TCP
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Events:
  Type    Reason            Age    From                   Message
  ----    ------            ----   ----                   -------
  Normal  SuccessfulCreate  7m39s  replicaset-controller  Created pod: hello-deploy-65cbc9474c-plhwm
  Normal  SuccessfulCreate  7m39s  replicaset-controller  Created pod: hello-deploy-65cbc9474c-8fdqs
  Normal  SuccessfulCreate  7m39s  replicaset-controller  Created pod: hello-deploy-65cbc9474c-dlv5g
  Normal  SuccessfulCreate  7m39s  replicaset-controller  Created pod: hello-deploy-65cbc9474c-x2ddb
  Normal  SuccessfulCreate  7m39s  replicaset-controller  Created pod: hello-deploy-65cbc9474c-qgffd
  Normal  SuccessfulCreate  7m39s  replicaset-controller  Created pod: hello-deploy-65cbc9474c-67nlw
  Normal  SuccessfulCreate  7m39s  replicaset-controller  Created pod: hello-deploy-65cbc9474c-gg4r2
  Normal  SuccessfulCreate  7m39s  replicaset-controller  Created pod: hello-deploy-65cbc9474c-b2v4v
  Normal  SuccessfulCreate  7m39s  replicaset-controller  Created pod: hello-deploy-65cbc9474c-vbzrd
  Normal  SuccessfulCreate  7m39s  replicaset-controller  (combined from similar events): Created pod: hello-deploy-65cbc9474c-pjwsh

访问部署的应用

对于Pod来说，其IP会由于销毁或者重建而改变，Pod的IP是一个变量，通常不能直接使用Pod的IP作为通讯。在K8s中，Kubernetes Service对象则解决了这一问题，它为一组Pod提供了一个固定的DNS域名和IP地址。

下面定义一个Service与上述Pod进行协同工作。

svc.yaml：

apiVersion: v1
kind: Service
metadata:
  name: hello-svc
  labels:
    app: hello-world
spec:
  type: NodePort
  ports:
  - port: 8080
    nodePort: 30001
    protocol: TCP
  selector:
    app: hello-world

部署Service：

root@lzl:/home/lzl# kubectl apply -f ./WorkSpace/k8s/service/svc.yaml
service/hello-svc created

这样我们有两种方式可以访问该应用：

在集群内部使用DNS名称hello-svc和端口8080访问
在集群外部通过集群的任意节点和端口号30001访问

首先我们查看我们刚部署的这10个Pod实例的具体信息：

root@lzl:/home/lzl# kubectl get pods -l app=hello-world -o wide
NAME                            READY   STATUS    RESTARTS   AGE   IP         NODE    NOMINATED NODE   READINESS GATES
hello-deploy-65cbc9474c-67nlw   1/1     Running   0          72m   10.5.1.4   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-8fdqs   1/1     Running   0          72m   10.5.2.6   lzl-c   <none>           <none>
hello-deploy-65cbc9474c-b2v4v   1/1     Running   0          72m   10.5.2.4   lzl-c   <none>           <none>
hello-deploy-65cbc9474c-dlv5g   1/1     Running   0          72m   10.5.2.7   lzl-c   <none>           <none>
hello-deploy-65cbc9474c-gg4r2   1/1     Running   0          72m   10.5.2.5   lzl-c   <none>           <none>
hello-deploy-65cbc9474c-pjwsh   1/1     Running   0          72m   10.5.2.3   lzl-c   <none>           <none>
hello-deploy-65cbc9474c-plhwm   1/1     Running   0          72m   10.5.1.5   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-qgffd   1/1     Running   0          72m   10.5.1.2   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-vbzrd   1/1     Running   0          72m   10.5.1.6   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-x2ddb   1/1     Running   0          72m   10.5.1.3   lzl-b   <none>           <none>

k8s自动为这些Pod分配了IP地址。

引自官方文档：

Kubernetes Service 从逻辑上定义了运行在集群中的一组 Pod，这些 Pod 提供了相同的功能。当每个 Service 创建时，会被分配一个唯一的 IP 地址（也称为 clusterIP）。这个 IP 地址与一个 Service 的生命周期绑定在一起，当 Service 存在的时候它也不会改变。可以配置 Pod 使它与 Service 进行通信，Pod 知道与 Service 通信将被自动地负载均衡到该 Service 中的某些 Pod 上。

我们查看创建的Service的信息：

root@lzl:/home/lzl# kubectl get svc hello-svc
NAME        TYPE       CLUSTER-IP       EXTERNAL-IP   PORT(S)          AGE
hello-svc   NodePort   10.110.196.234   <none>        8080:30001/TCP   19m
root@lzl:/home/lzl# kubectl describe svc hello-svc
Name:                     hello-svc
Namespace:                default
Labels:                   app=hello-world
Annotations:              <none>
Selector:                 app=hello-world
Type:                     NodePort
IP Family Policy:         SingleStack
IP Families:              IPv4
IP:                       10.110.196.234
IPs:                      10.110.196.234
Port:                     <unset>  8080/TCP
TargetPort:               8080/TCP
NodePort:                 <unset>  30001/TCP
Endpoints:                10.5.1.2:8080,10.5.1.3:8080,10.5.1.4:8080 + 7 more...
Session Affinity:         None
External Traffic Policy:  Cluster
Events:                   <none>

这里我遇到了集群内部和集群外部都无法访问应用的情况

集群内部无法访问
- 在Pod运行的节点上可以访问，但是在集群其他节点上无法访问
- 无法访问Service
集群外部无法访问：可以通过运行Pod的主机地址访问，而无法通过没有运行Pod的集群主机地址进行访问

解决方案在下一节

将问题解决后，我们来看。

root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get pods -o wide
NAME                           READY   STATUS    RESTARTS        AGE   IP          NODE    NOMINATED NODE   READINESS GATES
hello-deploy-59866ff45-2hp8j   1/1     Running   1 (8m20s ago)   39m   10.5.1.4    lzl-b   <none>           <none>
hello-deploy-59866ff45-4lhld   1/1     Running   1 (8m20s ago)   38m   10.5.1.11   lzl-b   <none>           <none>
hello-deploy-59866ff45-7pn94   1/1     Running   1 (8m20s ago)   35m   10.5.1.5    lzl-b   <none>           <none>
hello-deploy-59866ff45-9d444   1/1     Running   1 (8m20s ago)   35m   10.5.1.3    lzl-b   <none>           <none>
hello-deploy-59866ff45-lhx2j   1/1     Running   1 (8m20s ago)   37m   10.5.1.9    lzl-b   <none>           <none>
hello-deploy-59866ff45-nwcp8   1/1     Running   1 (8m20s ago)   36m   10.5.1.2    lzl-b   <none>           <none>
hello-deploy-59866ff45-prwwd   1/1     Running   1 (8m19s ago)   36m   10.5.1.7    lzl-b   <none>           <none>
hello-deploy-59866ff45-rflnv   1/1     Running   1 (8m19s ago)   34m   10.5.1.6    lzl-b   <none>           <none>
hello-deploy-59866ff45-tbpt2   1/1     Running   1 (8m19s ago)   37m   10.5.1.10   lzl-b   <none>           <none>
hello-deploy-59866ff45-z88ct   1/1     Running   1 (8m19s ago)   39m   10.5.1.8    lzl-b   <none>           <none>

root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get svc -o wide
NAME         TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)          AGE   SELECTOR
hello-svc    NodePort    10.101.13.201   <none>        8080:30001/TCP   18h   app=hello-world
kubernetes   ClusterIP   10.96.0.1       <none>        443/TCP          18h   <none>

从集群内访问

以服务实例hello-deploy-59866ff45-2hp8j为例，其地址为10.5.1.4:8080。因此，在集群内部任何一台机器访问该地址都可以访问到这个应用。

同样的，hello-svc的地址是10.101.13.201:8080，加入集群的任何一台主机都可以通过该地址访问到应用。

**特别的，与Pod的IP不同，Service的IP是固定不变的，而Pod的IP会跟随Deployment的更新、扩缩容等操作改变。**Service实际上相当于所有Pod的一个网关，它有固定的IP地址，并且将请求Pod的流量负载均衡到不同的Pod上，并在Pod地址改变时完成服务发现。

从集群外访问

如果在集群外，那么访问集群内的任何一台主机的30001端口都可以连接到该应用，因为前面svc.yaml：

 ports:
  - port: 8080
    nodePort: 30001
    protocol: TCP

帮我们做了集群内到集群外的端口映射。

解决由于iptables版本差异导致的K8s网络问题

如上文，我是用Ubuntu18.04搭建K8s1.22.4集群，在启动Service做应用网络入口时，出现了上述问题。找了很久的资料之后认为是k8s1.22.4在使用iptables配置网络规则时由于版本较新，其使用的配置命令在Ubuntu18.04版本上的iptables无法解析，导致网络出现问题。

https://bugs.launchpad.net/ubuntu/+source/linux-meta-hwe-5.4/+bug/1899690

所以我将虚拟机换为Ubuntu20.04后，该问题得到解决。由此看出，K8s对于环境版本一致性的要求还是比较高的。

解决由于虚拟机挂起，恢复虚拟机后集群网络异常的问题

在虚拟机挂起，再恢复之后，集群网络出现了问题，具体表现在：

无法通过Pod地址访问应用
无法通过Service地址访问应用
集群间Pod无法联通

在一篇博文中找到了解决方法：https://blog.csdn.net/weixin_43293361/article/details/114731838

首先，查看Master Node的网络，发现原本的cni0和flannel.1网络均不存在了：

然后查看了Work Node的网络，上述两个网络也是不存在的。所以按照博文给出的方法：

1）在Master Node删除flannel网络组件：

kubectl delete -f https://raw.githubusercontent.com/coreos/flannel/master/Documentation/kube-flannel.yml

2）在Master Node和Work Node上重置集群网络配置（集群所有节点上）：

ifconfig cni0 down
ip link delete cni0
ifconfig flannel.1 down
ip link delete flannel.1
rm -rf /var/lib/cni/flannel/*
rm -rf /var/lib/cni/networks/cbr0/*
rm -rf /var/lib/cni/cache/*
rm -f /etc/cni/net.d/*
systemctl restart kubelet
systemctl restart docker
chmod a+w /var/run/docker.sock

3）最后在Master Node重新安装flannel网络组件：

kubectl apply -f https://raw.githubusercontent.com/coreos/flannel/master/Documentation/kube-flannel.yml

4）查看集群各个节点的网络情况，cni0和flannel.1网络均正常出现，然后等待Deployment的Pod自愈，这样集群网络就可以正常恢复。

滚动升级与回滚

滚动升级

在升级使用的deploy.yaml文件中：

apiVersion: apps/v1
kind: Deployment
metadata:
  name: hello-deploy
spec:
  replicas: 10
  selector:
    matchLabels:
      app: hello-world
  minReadySeconds: 10
  strategy:
    type: RollingUpdate	# 滚动更新
    rollingUpdate:
      maxUnavailable: 1	# 集群中最多有一个不可用Pod
      maxSurge: 1		# 集群中Pod最多超过预期值1个
  template:
    metadata:
      labels:
        app: hello-world
    spec:
      containers:
      - name: hello-pod
        image: nigelpoulton/k8sbook:edge	# 使用了另一个版本的镜像
        ports:
        - containerPort: 8080

与上一个deploy.yaml不同，这里使用了另一个版本的镜像，更新策略由strategy下的内容决定。

进行滚动升级：

root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl apply -f deploy.yaml --record
Flag --record has been deprecated, --record will be removed in the future
deployment.apps/hello-deploy configured
root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl rollout status deployment hello-deploy
Waiting for deployment "hello-deploy" rollout to finish: 2 out of 10 new replicas have been updated...
^Croot@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get deploy
NAME           READY   UP-TO-DATE   AVAILABLE   AGE
hello-deploy   10/10   2            9           17h
root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get deploy
NAME           READY   UP-TO-DATE   AVAILABLE   AGE
hello-deploy   10/10   3            9           17h
root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get deploy
NAME           READY   UP-TO-DATE   AVAILABLE   AGE
hello-deploy   9/10    5            9           17h

可以看到，如Docker Swarm一样，在升级过程中，同时存在新版本与旧版本的应用，直至整个升级过程结束。

root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get deploy hello-deploy
NAME           READY   UP-TO-DATE   AVAILABLE   AGE
hello-deploy   10/10   10           10          17h

回滚

Kubernetes会维护Deployment的版本历史记录：

root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl rollout history deployment hello-deploy
deployment.apps/hello-deploy 
REVISION  CHANGE-CAUSE
1         <none>
2         kubectl apply --filename=deploy.yaml --record=true

对于Deployment依赖的ReplicaSet来说，每次更新都会保留旧版本的ReplicaSet同时创建一个新的ReplicaSet：

root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get rs
NAME                      DESIRED   CURRENT   READY   AGE
hello-deploy-59866ff45    10        10        10      158m
hello-deploy-65cbc9474c   0         0         0       20h
root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl describe rs
Name:           hello-deploy-59866ff45
Namespace:      default
Selector:       app=hello-world,pod-template-hash=59866ff45
Labels:         app=hello-world
                pod-template-hash=59866ff45
Annotations:    deployment.kubernetes.io/desired-replicas: 10
                deployment.kubernetes.io/max-replicas: 11
                deployment.kubernetes.io/revision: 2
                kubernetes.io/change-cause: kubectl apply --filename=deploy.yaml --record=true
Controlled By:  Deployment/hello-deploy
Replicas:       10 current / 10 desired
Pods Status:    10 Running / 0 Waiting / 0 Succeeded / 0 Failed
Pod Template:
  Labels:  app=hello-world
           pod-template-hash=59866ff45
  Containers:
   hello-pod:
    Image:        nigelpoulton/k8sbook:edge
    Port:         8080/TCP
    Host Port:    0/TCP
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Events:           <none>


Name:           hello-deploy-65cbc9474c
Namespace:      default
Selector:       app=hello-world,pod-template-hash=65cbc9474c
Labels:         app=hello-world
                pod-template-hash=65cbc9474c
Annotations:    deployment.kubernetes.io/desired-replicas: 10
                deployment.kubernetes.io/max-replicas: 11
                deployment.kubernetes.io/revision: 1
Controlled By:  Deployment/hello-deploy
Replicas:       0 current / 0 desired
Pods Status:    0 Running / 0 Waiting / 0 Succeeded / 0 Failed
Pod Template:
  Labels:  app=hello-world
           pod-template-hash=65cbc9474c
  Containers:
   hello-pod:
    Image:        nigelpoulton/k8sbook:latest
    Port:         8080/TCP
    Host Port:    0/TCP
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Events:           <none>

通过ReplicaSet的信息，可以清楚的知道两个ReplicaSet所做的操作，并且由于旧版本ReplicaSet的存在，回滚操作变得更加简单。下面将应用回滚到版本1，同时别忘记再更新一下yaml文件：

root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl rollout undo deployment hello-deploy --to-revision=1
deployment.apps/hello-deploy rolled back
root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get pods -o wide
NAME                            READY   STATUS              RESTARTS       AGE    IP          NODE    NOMINATED NODE   READINESS GATES
hello-deploy-59866ff45-2hp8j    1/1     Running             1 (135m ago)   166m   10.5.1.4    lzl-b   <none>           <none>
hello-deploy-59866ff45-4lhld    1/1     Running             1 (135m ago)   165m   10.5.1.11   lzl-b   <none>           <none>
hello-deploy-59866ff45-7pn94    1/1     Running             1 (135m ago)   162m   10.5.1.5    lzl-b   <none>           <none>
hello-deploy-59866ff45-9d444    1/1     Running             1 (135m ago)   162m   10.5.1.3    lzl-b   <none>           <none>
hello-deploy-59866ff45-lhx2j    1/1     Terminating         1 (135m ago)   164m   10.5.1.9    lzl-b   <none>           <none>
hello-deploy-59866ff45-nwcp8    1/1     Running             1 (135m ago)   163m   10.5.1.2    lzl-b   <none>           <none>
hello-deploy-59866ff45-prwwd    1/1     Running             1 (135m ago)   163m   10.5.1.7    lzl-b   <none>           <none>
hello-deploy-59866ff45-rflnv    1/1     Running             1 (135m ago)   161m   10.5.1.6    lzl-b   <none>           <none>
hello-deploy-59866ff45-tbpt2    1/1     Running             1 (135m ago)   165m   10.5.1.10   lzl-b   <none>           <none>
hello-deploy-59866ff45-z88ct    1/1     Running             1 (135m ago)   166m   10.5.1.8    lzl-b   <none>           <none>
hello-deploy-65cbc9474c-98bh2   0/1     ContainerCreating   0              10s    <none>      lzl-b   <none>           <none>
hello-deploy-65cbc9474c-d4g24   0/1     ContainerCreating   0              10s    <none>      lzl-b   <none>           <none>
root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl rollout status deployment hello-deploy
Waiting for deployment "hello-deploy" rollout to finish: 2 out of 10 new replicas have been updated...
^Croot@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get pods -o wide
NAME                            READY   STATUS              RESTARTS       AGE    IP          NODE    NOMINATED NODE   READINESS GATES
hello-deploy-59866ff45-2hp8j    1/1     Running             1 (136m ago)   167m   10.5.1.4    lzl-b   <none>           <none>
hello-deploy-59866ff45-4lhld    1/1     Running             1 (136m ago)   166m   10.5.1.11   lzl-b   <none>           <none>
hello-deploy-59866ff45-7pn94    1/1     Running             1 (136m ago)   163m   10.5.1.5    lzl-b   <none>           <none>
hello-deploy-59866ff45-9d444    1/1     Running             1 (136m ago)   162m   10.5.1.3    lzl-b   <none>           <none>
hello-deploy-59866ff45-nwcp8    1/1     Running             1 (136m ago)   164m   10.5.1.2    lzl-b   <none>           <none>
hello-deploy-59866ff45-prwwd    1/1     Running             1 (136m ago)   164m   10.5.1.7    lzl-b   <none>           <none>
hello-deploy-59866ff45-rflnv    1/1     Terminating         1 (136m ago)   162m   10.5.1.6    lzl-b   <none>           <none>
hello-deploy-59866ff45-tbpt2    1/1     Running             1 (136m ago)   165m   10.5.1.10   lzl-b   <none>           <none>
hello-deploy-59866ff45-z88ct    1/1     Running             1 (136m ago)   167m   10.5.1.8    lzl-b   <none>           <none>
hello-deploy-65cbc9474c-98bh2   1/1     Running             0              47s    10.5.1.12   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-d4g24   0/1     ContainerCreating   0              47s    <none>      lzl-b   <none>           <none>
hello-deploy-65cbc9474c-kglhr   0/1     ContainerCreating   0              2s     <none>      lzl-b   <none>           <none>
root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl rollout status deployment hello-deploy
Waiting for deployment "hello-deploy" rollout to finish: 3 out of 10 new replicas have been updated...
^[[AWaiting for deployment "hello-deploy" rollout to finish: 3 out of 10 new replicas have been updated...
^Croot@lzl-a:/home/lzl/WP/k8s/deploykubectl get deploy hello-deploy
NAME           READY   UP-TO-DATE   AVAILABLE   AGE
hello-deploy   10/10   3            9           20h
root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get deploy hello-deploy
NAME           READY   UP-TO-DATE   AVAILABLE   AGE
hello-deploy   10/10   10           10          20h
root@lzl-a:/home/lzl/WP/k8s/deploy# kubectl get pods -o wide
NAME                            READY   STATUS    RESTARTS   AGE     IP          NODE    NOMINATED NODE   READINESS GATES
hello-deploy-65cbc9474c-7jrwq   1/1     Running   0          4m32s   10.5.1.19   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-98bh2   1/1     Running   0          7m56s   10.5.1.12   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-d4g24   1/1     Running   0          7m56s   10.5.1.13   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-hkr5r   1/1     Running   0          6m39s   10.5.1.15   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-jdwsr   1/1     Running   0          3m29s   10.5.1.21   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-kglhr   1/1     Running   0          7m11s   10.5.1.14   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-n4mqz   1/1     Running   0          4m1s    10.5.1.20   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-p7zgs   1/1     Running   0          6m7s    10.5.1.16   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-rqtrg   1/1     Running   0          5m36s   10.5.1.17   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-xxk2n   1/1     Running   0          5m4s    10.5.1.18   lzl-b   <none>           <none>

和滚动升级一样，回滚过程中也同时存在新旧版本的应用。由于回滚是命令操作，没有使用到deploy.yaml文件，因此要注意应用版本与部署文件中镜像版本不一致的问题，可以手动更改部署文件。

删除Deployment和Service

不再使用该Service和Pod时，将这两个对象删除，使用命令：kubectl delete -f。

root@lzl-a:/home/lzl/WP/k8s# kubectl delete -f ./deploy/deploy.yaml 
deployment.apps "hello-deploy" deleted
root@lzl-a:/home/lzl/WP/k8s# kubectl delete -f ./service/svc.yaml 
service "hello-svc" deleted
root@lzl-a:/home/lzl/WP/k8s# kubectl get pods -o wide
NAME                            READY   STATUS        RESTARTS   AGE     IP          NODE    NOMINATED NODE   READINESS GATES
hello-deploy-65cbc9474c-7jrwq   1/1     Terminating   0          10m     10.5.1.19   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-98bh2   1/1     Terminating   0          13m     10.5.1.12   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-d4g24   1/1     Terminating   0          13m     10.5.1.13   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-hkr5r   1/1     Terminating   0          12m     10.5.1.15   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-jdwsr   1/1     Terminating   0          9m19s   10.5.1.21   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-kglhr   1/1     Terminating   0          13m     10.5.1.14   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-n4mqz   1/1     Terminating   0          9m51s   10.5.1.20   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-p7zgs   1/1     Terminating   0          11m     10.5.1.16   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-rqtrg   1/1     Terminating   0          11m     10.5.1.17   lzl-b   <none>           <none>
hello-deploy-65cbc9474c-xxk2n   1/1     Terminating   0          10m     10.5.1.18   lzl-b   <none>           <none>
root@lzl-a:/home/lzl/WP/k8s# kubectl get svc -o wide
NAME         TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)   AGE   SELECTOR
kubernetes   ClusterIP   10.96.0.1    <none>        443/TCP   20h   <none>
root@lzl-a:/home/lzl/WP/k8s# kubectl get pods -o wide
No resources found in default namespace.