Kubernetes学习之路（六）之创建K8S应用

• 一、Deployment的概念

K8S本身并不提供网络的功能，所以需要借助第三方网络插件进行部署K8S中的网络，以打通各个节点中容器的互通。 
POD，是K8S中的一个逻辑概念，K8S管理的是POD，一个POD中包含多个容器，容器之间通过localhost互通。而POD需要ip地址。每个POD都有一个标签

POD–>RC–>RS–>Deployment （发展历程）

Deployment，表示用户对K8S集群的一次更新操作。Deployment是一个比RS应用模式更广的API对象。用于保证Pod的副本的数量。

可以是创建一个新的服务，更新一个新的服务，也可以是滚动升级一个服务。滚动升级一个服务。实际是创建一个新的RS，然后将新的RS中副本数增加到理想状态，将旧的RS中的副本数减小到0的复合操作； 这样的一个复合操作用一个RS是不太好描述的，所以用一个更通用的Deployment来描述。

RC、RS和Deployment只是保证了支撑服务的POD数量，但是没有解决如何访问这些服务的问题。一个POD只是一个运行服务的实例，随时可以能在一个节点上停止，在另一个节点以一个新的IP启动一个新的POD，因此不能以确定的IP和端口号提供服务。

要稳定地提供服务需要服务发现和负载均衡能力。服务发现完成的工作，是针对客户端访问的服务，找到对应的后端服务实例。

在K8S的集中当中，客户端需要访问的服务就是Service对象。每个Service会对应一个集群内部有效的虚拟IP，集群内部通过虚拟IP访问一个服务。

• 二、创建K8S的第一个应用

[root@linux-node1 ~]# kubectl run net-test --image=alpine --replicas= sleep 　　#创建名称为net-test的应用，镜像指定为alpine，副本数为2个
deployment.apps "net-test" created
[root@linux-node1 ~]# kubectl get pod -o wide　　#查看pod的状态信息，此时是API Server从etcd中读取这些数据
NAME                        READY     STATUS              RESTARTS   AGE       IP          NODE
net-test-7b949fc785-2v2qz   /       Running                       56s       10.2.87.2   192.168.56.120
net-test-7b949fc785-6nrhm   /       ContainerCreating             56s       <none>      192.168.56.130
[root@linux-node1 ~]# kubectl get deployment net-test
NAME       DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
net-test                                            22h

kubectl get deployment命令可以查看net-test的状态，输出显示两个副本正常运行。还可以在创建的过程中，通过kubectl describe deployment net-test了解详细的信息。  

[root@linux-node1 ~]# kubectl describe deployment net-test
Name:                   net-test
Namespace:              default
CreationTimestamp:      Thu,  Aug  :: +
Labels:                 run=net-test
Annotations:            deployment.kubernetes.io/revision=
Selector:               run=net-test
Replicas:                desired |  updated |  total |  available |  unavailable
StrategyType:           RollingUpdate
MinReadySeconds:
RollingUpdateStrategy:   max unavailable,  max surge
Pod Template:
  Labels:  run=net-test
  Containers:
   net-test:
    Image:      alpine
    Port:       <none>
    Host Port:  <none>
    Args:
      sleep

    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Conditions:
  Type           Status  Reason
  ----           ------  ------
  Available      True    MinimumReplicasAvailable
  Progressing    True    NewReplicaSetAvailable
OldReplicaSets:  <none>
NewReplicaSet:   net-test-5767cb94df (/ replicas created)
Events:          <none>　　

Events是Deployment的日志，记录整个RelicaSet的启动过程，从上面的创建过程，可以看到Deployment是通过ReplicaSet来管理Pod。

[root@linux-node1 ~]# kubectl get replicaset　　#获取副本集信息
NAME                          DESIRED   CURRENT   READY     AGE
net-test-5767cb94df                                      23h

[root@linux-node1 ~]# kubectl describe replicaset net-test-5767cb94df　　#查看副本集的详细信息
Name:           net-test-5767cb94df
Namespace:      default
Selector:       pod-template-hash=,run=net-test
Labels:         pod-template-hash=
                run=net-test
Annotations:    deployment.kubernetes.io/desired-replicas=
                deployment.kubernetes.io/max-replicas=
                deployment.kubernetes.io/revision=
Controlled By:  Deployment/net-test　　#指明ReplicaSet是由Deployment net-test创建
Replicas:        current /  desired
Pods Status:     Running /  Waiting /  Succeeded /  Failed
Pod Template:
  Labels:  pod-template-hash=
           run=net-test
  Containers:
   net-test:
    Image:      alpine
    Port:       <none>
    Host Port:  <none>
    Args:
      sleep

    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Events:           <none>　　#Events可以查看到两个副本Pod的创建过程

[root@linux-node1 ~]# kubectl get pod　　#获取Pod信息，可以看到2个副本都处于Running状态
NAME                                READY     STATUS    RESTARTS   AGE
net-test-5767cb94df-djt98           /       Running             22h
net-test-5767cb94df-zb8m4           /       Running             23h

[root@linux-node1 ~]# kubectl describe pod net-test-5767cb94df-djt98 #查看pod的详细信息
Name:           net-test-5767cb94df-djt98
Namespace:      default
Node:           192.168.56.13/192.168.56.13
Start Time:     Thu,  Aug  :: +
Labels:         pod-template-hash=
                run=net-test
Annotations:    <none>
Status:         Running
IP:             10.2.73.3
Controlled By:  ReplicaSet/net-test-5767cb94df
Containers:
  net-test:
    Container ID:  docker://c8e267326ed80f3cbe8111377c74dd1f016beaef513196b941165e180a5d5733
    Image:         alpine
    Image ID:      docker-pullable://alpine@sha256:7043076348bf5040220df6ad703798fd8593a0918d06d3ce30c6c93be117e430
    Port:          <none>
    Host Port:     <none>
    Args:
      sleep

    State:          Running
      Started:      Thu,  Aug  :: +
    Ready:          True
    Restart Count:
    Environment:    <none>
    Mounts:
      /var/run/secrets/kubernetes.io/serviceaccount from default-token-mnqx5 (ro)
Conditions:
  Type           Status
  Initialized    True
  Ready          True
  PodScheduled   True
Volumes:
  default-token-mnqx5:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  default-token-mnqx5
    Optional:    false
QoS Class:       BestEffort
Node-Selectors:  <none>
Tolerations:     <none>
Events:          <none>

Controlled By 指明此 Pod 是由 ReplicaSet/net-test-5767cb94df 创建。Events 记录了 Pod 的启动过程。如果操作失败（比如 image 不存在），也能在这里查看到原因。

总结创建的过程：

（1）用户通过kubectl创建Deployment

（2）Deployment创建ReplicaSet

（3）ReplicaSet创建Pod

如图：

• 三、K8S创建资源的两种方式

Kubernetes 支持两种方式创建资源：

（1）用kubectl命令直接创建，在命令行中通过参数指定资源的属性。此方式简单直观，比较适合临时测试或实验使用。

kubectl run net-test --image=alpine --replicas= sleep 

（2）通过配置文件和kubectl create创建。在配置文件中描述了应用的信息和需要达到的预期状态。

kubectl create -f nginx-deployment.yaml

• 四、以Deployment YAML方式创建Nginx服务

• 1、创建deployment

[root@linux-node1 ~]# vim nginx-deployment.yaml　　#使用yaml的方式进行创建应用
apiVersion: apps/v1　　#apiVersion是当前配置格式的版本
kind: Deployment　　　　#kind是要创建的资源类型，这里是Deploymnet
metadata:　　　　　　　　#metadata是该资源的元数据，name是必须的元数据项
  name: nginx-deployment
  labels:
    app: nginx
spec:　　　　　　　　　　#spec部分是该Deployment的规则说明
  replicas: 3　　　　　 #relicas指定副本数量，默认为1
  selector:
    matchLabels:
      app: nginx
  template:　　　　　　#template定义Pod的模板，这是配置的重要部分
    metadata:　　　　  #metadata定义Pod的元数据，至少要顶一个label，label的key和value可以任意指定
      labels:
        app: nginx
    spec:　　　　　　　#spec描述的是Pod的规则，此部分定义pod中每一个容器的属性，name和image是必需的
      containers:
      - name: nginx
        image: nginx:1.13.
        ports:
        - containerPort: 

[root@linux-node1 ~]# kubectl create -f nginx-deployment.yaml #创建nginx-deployment应用
deployment.apps "nginx-deployment" created

• 2、查看deployment

[root@linux-node1 ~]# kubectl get deployment
NAME               DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
net-test                                                    32m
nginx-deployment   3         3         3            0           10s

[root@linux-node1 ~]# kubectl describe deployment nginx-deployment    #查看deployment详情
Name:                   nginx-deployment
Namespace:              default
CreationTimestamp:      Thu,  Aug  :: +
Labels:                 app=nginx
Annotations:            deployment.kubernetes.io/revision=
Selector:               app=nginx
Replicas:                desired |  updated |  total |  available |  unavailable
StrategyType:           RollingUpdate
MinReadySeconds:
RollingUpdateStrategy:  % max unavailable, % max surge
Pod Template:
  Labels:  app=nginx
  Containers:
   nginx:
    Image:        nginx:1.13.
    Port:         /TCP
    Host Port:    /TCP
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Conditions:
  Type           Status  Reason
  ----           ------  ------
  Available      False   MinimumReplicasUnavailable
  Progressing    True    ReplicaSetUpdated
OldReplicaSets:  <none>
NewReplicaSet:   nginx-deployment-6c45fc49cb (/ replicas created)
Events:
  Type    Reason             Age   From                   Message
  ----    ------             ----  ----                   -------
  Normal  ScalingReplicaSet  1m    deployment-controller  Scaled up replica set nginx-deployment-6c45fc49cb to 

• 3、查看Pod

[root@linux-node1 ~]# kubectl get pod    #查看pod在状态，正在创建中，此时应该正在拉取镜像
NAME                                READY     STATUS              RESTARTS   AGE
net-test-5767cb94df-djt98           /       Running                       22m
net-test-5767cb94df-hcwv7           /       Unknown                       34m
net-test-5767cb94df-zb8m4           /       Running                       34m
nginx-deployment-6c45fc49cb-dmc22   /       ContainerCreating             2m
nginx-deployment-6c45fc49cb-fd8xm   /       ContainerCreating             2m
nginx-deployment-6c45fc49cb-sc8sh   /       ContainerCreating             2m

[root@linux-node1 ~]# kubectl describe pod nginx-deployment-6c45fc49cb-dmc22  #查看具体某个pod的状态信息

[root@linux-node1 ~]# kubectl get pod -o wide　　#创建成功，状态为Running
NAME                                READY     STATUS    RESTARTS   AGE       IP          NODE
net-test-5767cb94df-djt98           /       Running             24m       10.2.73.3   192.168.56.13
net-test-5767cb94df-hcwv7           /       Unknown             36m       10.2.10.2   192.168.56.12
net-test-5767cb94df-zb8m4           /       Running             36m       10.2.73.2   192.168.56.13
nginx-deployment-6c45fc49cb-dmc22   1/1       Running   0          4m        10.2.73.6   192.168.56.13
nginx-deployment-6c45fc49cb-fd8xm   1/1       Running   0          4m        10.2.73.4   192.168.56.13
nginx-deployment-6c45fc49cb-sc8sh   1/1       Running   0          4m        10.2.73.5   192.168.56.13

Deployment、ReplicaSet、Pod 都已经就绪。如果要删除这些资源，执行 kubectl delete deployment nginx-deployment 或者 kubectl delete -f nginx-deployment.yaml。

• 4、测试Pod访问

[root@linux-node1 ~]# curl --head http://10.2.73.6
HTTP/1.1  OK
Server: nginx/1.13.12
Date: Thu,  Aug  :: GMT
Content-Type: text/html
Content-Length:
Last-Modified: Mon,  Apr  :: GMT
Connection: keep-alive
ETag: "5acb8e45-264"
Accept-Ranges: bytes

• 5、更新Deployment

[root@linux-node1 ~]# kubectl set image deployment/nginx-deployment nginx=nginx:1.15. --record    #nginx的版本升级，由1.13.2升级为1.15.2，记录需要加参数--record
deployment.apps "nginx-deployment" image updated

[root@linux-node1 ~]# kubectl get deployment -o wide　　#查看更新后的deployment，可以看到当前4个副本，说明还在滚动升级中
NAME               DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE       CONTAINERS   IMAGES         SELECTOR
net-test                                                    39m       net-test     alpine         run=net-test
nginx-deployment                                            6m        nginx        nginx:1.15.2   app=nginx

• 6、查看更新历史

[root@linux-node1 ~]# kubectl rollout history deployment/nginx-deployment #查看更新历史记录
deployments "nginx-deployment"
REVISION  CHANGE-CAUSE
         <none>
         kubectl set image deployment/nginx-deployment nginx=nginx:1.15. --record=true

• 7、查看具体某一个版本的升级历史

[root@linux-node1 ~]# kubectl rollout history deployment/nginx-deployment --revision=
deployments "nginx-deployment" with revision #
Pod Template:
  Labels:    app=nginx
    pod-template-hash=
  Containers:
   nginx:
    Image:    nginx:1.13.
    Port:    /TCP
    Host Port:    /TCP
    Environment:    <none>
    Mounts:    <none>
  Volumes:    <none>

• 8、查看更新后的Deployment，并进行访问

[root@linux-node1 ~]# kubectl get pod -o wide
NAME                                READY     STATUS    RESTARTS   AGE       IP          NODE
net-test-5767cb94df-djt98           /       Running             30m       10.2.73.3   192.168.56.13
net-test-5767cb94df-hcwv7           /       Unknown             42m       10.2.10.2   192.168.56.12
net-test-5767cb94df-zb8m4           /       Running             42m       10.2.73.2   192.168.56.13
nginx-deployment-64749d4b59-djttr   1/1       Running   0          37s       10.2.73.8   192.168.56.13
nginx-deployment-64749d4b59-jp7fw   1/1       Running   0          3m        10.2.73.7   192.168.56.13
nginx-deployment-64749d4b59-q4fsn   1/1       Running   0          33s       10.2.73.9   192.168.56.13
[root@linux-node1 ~]# curl --head http://10.2.73.7
HTTP/1.1  OK
Server: nginx/1.15.2　　#版本已经升级为1.15.2
Date: Thu,  Aug  :: GMT
Content-Type: text/html
Content-Length:
Last-Modified: Tue,  Jul  :: GMT
Connection: keep-alive
ETag: "5b572365-264"
Accept-Ranges: bytes

• 9、快速回滚到上一个版本

[root@linux-node1 ~]# kubectl rollout undo deployment/nginx-deployment   #回滚上一个版本
deployment.apps "nginx-deployment"    

[root@linux-node1 ~]# kubectl get pod -o wide
NAME                                READY     STATUS    RESTARTS   AGE       IP           NODE
net-test-5767cb94df-djt98           /       Running             32m       10.2.73.3    192.168.56.13
net-test-5767cb94df-hcwv7           /       Unknown             43m       10.2.10.2    192.168.56.12
net-test-5767cb94df-zb8m4           /       Running             43m       10.2.73.2    192.168.56.13
nginx-deployment-6c45fc49cb-b9h84   /       Running             24s       10.2.73.11   192.168.56.13
nginx-deployment-6c45fc49cb-g4mrg   /       Running             26s       10.2.73.10   192.168.56.13
nginx-deployment-6c45fc49cb-k29kq   /       Running             21s       10.2.73.12   192.168.56.13
[root@linux-node1 ~]# curl --head http://10.2.73.10
HTTP/1.1  OK
Server: nginx/1.13.12
Date: Thu,  Aug  :: GMT
Content-Type: text/html
Content-Length:
Last-Modified: Mon,  Apr  :: GMT
Connection: keep-alive
ETag: "5acb8e45-264"
Accept-Ranges: bytes

回滚完成，每一次更新或者回滚ip都会变化，所以需要通过vip进行访问，这就引入了service   

• 10、使用service的vip进行访问应用

[root@linux-node1 ~]# vim nginx-service.yaml　　#使用yaml方式创建service
kind: Service
apiVersion: v1
metadata:
  name: nginx-service
spec:
  selector:
    app: nginx
  ports:
  - protocol: TCP
    port:
    targetPort: 

[root@linux-node1 ~]# kubectl create -f nginx-service.yaml 　　#创建service
service "nginx-service" created

[root@linux-node1 ~]# kubectl get service
NAME            TYPE        CLUSTER-IP     EXTERNAL-IP   PORT(S)   AGE
kubernetes      ClusterIP   10.1.0.1       <none>        /TCP   4h
nginx-service   ClusterIP   10.1.213.126   <none>        /TCP    15s　　#这个就是vip

[root@linux-node2 ~]# curl --head http://10.1.213.126　　#在node2节点上进行访问vip测试，在node1上无法访问是因为没有安装kube-proxy导致无法访问
HTTP/1.1  OK
Server: nginx/1.13.
Date: Thu,  Aug  :: GMT
Content-Type: text/html
Content-Length:
Last-Modified: Mon,  Apr  :: GMT
Connection: keep-alive
ETag: "5acb8e45-264"
Accept-Ranges: bytes

[root@linux-node2 ~]# ipvsadm -Ln　　
IP Virtual Server version 1.2. (size=)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
TCP  10.1.0.1: rr persistent
  -> 192.168.56.11:           Masq
TCP  10.1.213.126: rr
  -> 10.2.73.10:                Masq
  -> 10.2.73.11:                Masq
  -> 10.2.73.12:                Masq                       

查看LVS状态可以看到，当访问VIP：10.1.213.126时，会进行负载均衡到各个pod      

• 11、扩容到5个节点

[root@linux-node1 ~]# kubectl scale deployment nginx-deployment --replicas 5　　#对应用的副本数进行扩容，直接指定副本数为5
deployment.extensions "nginx-deployment" scaled

[root@linux-node1 ~]# kubectl get pod　　#查看pod状态，可以看到已经增加到5个副本
NAME                                READY     STATUS    RESTARTS   AGE
net-test-5767cb94df-djt98           /       Running             38m
net-test-5767cb94df-hcwv7           /       Unknown             50m
net-test-5767cb94df-zb8m4           /       Running             50m
nginx-deployment-6c45fc49cb-b9h84   /       Running             6m
nginx-deployment-6c45fc49cb-g4mrg   /       Running             7m
nginx-deployment-6c45fc49cb-k29kq   /       Running             6m
nginx-deployment-6c45fc49cb-n9qkx   /       Running             24s
nginx-deployment-6c45fc49cb-xpx9s   /       Running             24s

[root@linux-node2 ~]# ipvsadm -Ln
IP Virtual Server version 1.2. (size=)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
TCP  10.1.0.1: rr persistent
  -> 192.168.56.11:           Masq
TCP  10.1.213.126: rr
  -> 10.2.73.10:                Masq
  -> 10.2.73.11:                Masq
  -> 10.2.73.12:                Masq
  -> 10.2.73.13:                Masq
  -> 10.2.73.14:                Masq