Kubernetes 工作负载控制器Deployment和Replicaset

Kubernetes 工作负载控制器Deployment和Replicaset

Deployment：介绍

应用场景：网站、API、微服务

Deployment：使用流程

项目生命周期

Deployment：部署

第一步：部署镜像

• kubectl apply -f xxx.yaml

• kubectl create deployment web --image=nginx:1.15

[root@k8s-master ~]# cat deployment.yaml apiVersion: apps/v1kind: Deploymentmetadata: name: web namespace: defaultspec: replicas: 3 selector: matchLabels: app: nginx strategy: type: RollingUpdate rollingUpdate: maxSurge: 1 maxUnavailable: 2 template: metadata: labels: app: nginx spec: containers: - name: web image: nginx:1.16 ports: - containerPort: 80 env: - name: "VERSION" value: "1.19" livenessProbe: path: /index.html port: 80 readinessProbe: path: /index.html port: 80 initialDelaySeconds: 10 periodSeconds: 10 resources: requests: memory: "64Mi" cpu: "250m" limits: memory: "128Mi" cpu: "500m" [root@k8s-master ~]# kubectl apply -f deployment.yaml deployment.apps/web created[root@k8s-master ~]# kubectl get pod -o wideNAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATESweb-7cc58d88d9-7rvwq 1/1 Running 0 22m 10.244.169.149 k8s-node2 web-7cc58d88d9-b786z 1/1 Running 0 22m 10.244.169.148 k8s-node2 web-7cc58d88d9-xqdqx 1/1 Running 0 22m 10.244.36.82 k8s-node1

第二步：暴露服务

[root@k8s-master ~]# cat service.yaml apiVersion: v1kind: Servicemetadata: name: web namespace: defaultspec: ports: - port: 80 protocol: TCP targetPort: 80 selector: app: nginx type: NodePort [root@k8s-master ~]# kubectl apply -f service.yaml service/web created[root@k8s-master ~]# kubectl get svcNAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGEkubernetes ClusterIP 10.96.0.1 443/TCP 32dweb NodePort 10.96.100.183 80:30866/TCP 88s[root@k8s-master ~]# kubectl get epNAME ENDPOINTS AGEkubernetes 192.168.179.102:6443 32dweb 10.244.169.148:80,10.244.169.149:80,10.244.36.82:80 2m42s

现在可以根据任意node节点的IP+30866就可以访问到pod了

第三步 Deployment：滚动升级

（更新镜像三种方式）

• kubectl apply -f xxx.yaml

• kubectl set image deployment/web nginx=nginx:1.16

• kubectl edit deployment/webkubec

image: nginx:1.16

滚动更新查看replicaset状态

[root@k8s-master ~]# kubectl get replicaset -wNAME DESIRED CURRENT READY AGEweb-7cc58d88d9 3 3 3 32mweb-6c8f6b7f84 1 0 0 0sweb-7cc58d88d9 1 3 3 34mweb-6c8f6b7f84 1 0 0 0sweb-6c8f6b7f84 3 0 0 0sweb-7cc58d88d9 1 3 3 34mweb-6c8f6b7f84 3 1 0 1sweb-7cc58d88d9 1 1 1 34mweb-6c8f6b7f84 3 1 0 1sweb-6c8f6b7f84 3 3 0 1sweb-6c8f6b7f84 3 3 1 62sweb-7cc58d88d9 0 1 1 35mweb-7cc58d88d9 0 1 1 35mweb-7cc58d88d9 0 0 0 35mweb-6c8f6b7f84 3 3 2 62sweb-6c8f6b7f84 3 3 3 64s^C[root@k8s-master ~]# kubectl get replicasetNAME DESIRED CURRENT READY AGEweb-6c8f6b7f84 3 3 3 4m55sweb-7cc58d88d9 0 0 0 39m

滚动更新查看ep状态

[root@k8s-master ~]# kubectl get ep -wNAME ENDPOINTS AGEkubernetes 192.168.179.102:6443 32dweb 10.244.169.148:80,10.244.169.149:80,10.244.36.82:80 7m33sweb 10.244.169.149:80,10.244.36.82:80 9m2sweb 10.244.36.82:80 9m2sweb 10.244.36.82:80 9m47sweb 10.244.36.82:80 9m49sweb 10.244.36.82:80 9m52sweb 10.244.36.82:80,10.244.36.83:80 10mweb 10.244.36.83:80 10mweb 10.244.169.150:80,10.244.36.83:80 10mweb 10.244.169.150:80,10.244.169.151:80,10.244.36.83:80 10m[root@k8s-master ~]# kubectl get epNAME ENDPOINTS AGEkubernetes 192.168.179.102:6443 32dweb 10.244.169.150:80,10.244.169.151:80,10.244.36.83:80 15m

滚动更新查看pod状态（可以看到是有个逐步更新替代的过程，健康检查会影响pod更新速率，因为只有健康检查通过才认为pod是runing状态）

[root@k8s-master ~]# kubectl get pod -wNAME READY STATUS RESTARTS AGEweb-7cc58d88d9-7rvwq 1/1 Running 0 33mweb-7cc58d88d9-b786z 1/1 Running 0 33mweb-7cc58d88d9-xqdqx 1/1 Running 0 33mweb-6c8f6b7f84-tgqcj 0/1 Pending 0 0sweb-6c8f6b7f84-tgqcj 0/1 Pending 0 0sweb-7cc58d88d9-b786z 1/1 Terminating 0 34mweb-7cc58d88d9-7rvwq 1/1 Terminating 0 34mweb-6c8f6b7f84-tgqcj 0/1 ContainerCreating 0 1sweb-6c8f6b7f84-2kpvr 0/1 Pending 0 0sweb-6c8f6b7f84-2kpvr 0/1 Pending 0 0sweb-6c8f6b7f84-l55tq 0/1 Pending 0 0sweb-6c8f6b7f84-l55tq 0/1 Pending 0 0sweb-6c8f6b7f84-2kpvr 0/1 ContainerCreating 0 0sweb-7cc58d88d9-b786z 1/1 Terminating 0 34mweb-7cc58d88d9-7rvwq 1/1 Terminating 0 34mweb-6c8f6b7f84-2kpvr 0/1 ContainerCreating 0 2sweb-6c8f6b7f84-tgqcj 0/1 ContainerCreating 0 4sweb-7cc58d88d9-b786z 0/1 Terminating 0 34mweb-7cc58d88d9-7rvwq 0/1 Terminating 0 34mweb-7cc58d88d9-b786z 0/1 Terminating 0 34mweb-7cc58d88d9-b786z 0/1 Terminating 0 34mweb-7cc58d88d9-7rvwq 0/1 Terminating 0 34mweb-7cc58d88d9-7rvwq 0/1 Terminating 0 34mweb-6c8f6b7f84-l55tq 0/1 Pending 0 8sweb-6c8f6b7f84-l55tq 0/1 ContainerCreating 0 8sweb-6c8f6b7f84-l55tq 0/1 ContainerCreating 0 9sweb-6c8f6b7f84-2kpvr 0/1 Running 0 44sweb-6c8f6b7f84-tgqcj 0/1 Running 0 48sweb-6c8f6b7f84-l55tq 0/1 Running 0 50sweb-6c8f6b7f84-2kpvr 1/1 Running 0 61sweb-7cc58d88d9-xqdqx 1/1 Terminating 0 35mweb-6c8f6b7f84-tgqcj 1/1 Running 0 62sweb-7cc58d88d9-xqdqx 1/1 Terminating 0 35mweb-7cc58d88d9-xqdqx 0/1 Terminating 0 35mweb-6c8f6b7f84-l55tq 1/1 Running 0 63sweb-7cc58d88d9-xqdqx 0/1 Terminating 0 35mweb-7cc58d88d9-xqdqx 0/1 Terminating 0 35m

滚动升级： K8s对Pod升级的默认策略，通过使 用新版本Pod逐步更新旧版本Pod，实现零停机 发布，用户无感知。

滚动升级大致意思是先启动新pod，然后替换旧的pod

这里执行了Up down这个过程，这里用到replicaset，这个控制器我们平时一般不会去使用，这个控制器由deployment创建出来的，deployment就利用了replicaset来实现滚动更新机制

[root@k8s-master ~]# kubectl get replicasetNAME DESIRED CURRENT READY AGEweb-6c8f6b7f84 3 3 3 14mweb-7cc58d88d9 0 0 0 48m[root@k8s-master ~]# kubectl get deploymentNAME READY UP-TO-DATE AVAILABLE AGEweb 3/3 3 3 47m[root@k8s-master ~]# kubectl describe deployment webEvents: Type Reason Age From Message ---- ------ ---- ---- ------- Normal ScalingReplicaSet 48m deployment-controller Scaled up replica set web-7cc58d88d9 to 3 Normal ScalingReplicaSet 13m deployment-controller Scaled up replica set web-6c8f6b7f84 to 1 Normal ScalingReplicaSet 13m deployment-controller Scaled down replica set web-7cc58d88d9 to 1 Normal ScalingReplicaSet 13m deployment-controller Scaled up replica set web-6c8f6b7f84 to 3 Normal ScalingReplicaSet 12m deployment-controller Scaled down replica set web-7cc58d88d9 to 0

第四步：水平扩缩容（启动多实例，提高并发）

• 修改yaml里replicas值，再apply

• kubectl scale deployment web --replicas=10    注：replicas参数控制Pod副本数量

第四步：回滚（发布失败恢复正常版本）

kubectl rollout history deployment/web # 查看历史发布版本

kubectl rollout undo deployment/web # 回滚上一个版本

kubectl rollout undo deployment/web --to-revision=2 # 回滚历史指定版本

注：回滚是重新部署某一次部署时的状态，即当时版本所有配置

[root@k8s-master ~]# kubectl rollout history deployment/webdeployment.apps/web REVISION CHANGE-CAUSE1 2 [root@k8s-master ~]# kubectl rollout undo deployment/webdeployment.apps/web rolled back[root@k8s-master ~]# kubectl get replicaset -wNAME DESIRED CURRENT READY AGEweb-6c8f6b7f84 3 3 3 29mweb-7cc58d88d9 0 0 0 63mweb-7cc58d88d9 0 0 0 64mweb-7cc58d88d9 1 0 0 64mweb-6c8f6b7f84 1 3 3 30mweb-7cc58d88d9 1 0 0 64mweb-7cc58d88d9 3 0 0 64mweb-6c8f6b7f84 1 3 3 30mweb-6c8f6b7f84 1 1 1 30mweb-7cc58d88d9 3 1 0 64mweb-7cc58d88d9 3 1 0 64mweb-7cc58d88d9 3 3 0 64mweb-7cc58d88d9 3 3 1 64mweb-6c8f6b7f84 0 1 1 30mweb-6c8f6b7f84 0 1 1 30mweb-7cc58d88d9 3 3 2 64mweb-6c8f6b7f84 0 0 0 30mweb-7cc58d88d9 3 3 3 64m^C[root@k8s-master ~]# kubectl get replicasetNAME DESIRED CURRENT READY AGEweb-6c8f6b7f84 0 0 0 30mweb-7cc58d88d9 3 3 3 64m

第五步：Deployment：删除

最后，项目下线：

kubectl delete deploy/web

kubectl delete svc/web

Deployment：ReplicaSet

ReplicaSet控制器用途：

• Pod副本数量管理，不断对比当前Pod数量与期望Pod数量

• Deployment每次发布都会创建一个RS作为记录，用于实现回滚

kubectl get rs #查看RS记录

kubectl rollout history deployment web #版本对应RS记录

Replicaset后面有个死循环，一直判断当前的容器数量是不是预定义的，也就是管理pod副本的数量，记录每次deployment的每次发布版本，用于实现回滚，作为deployment回滚的一个记录。

[root@k8s-master ~]# kubectl get rsNAME DESIRED CURRENT READY AGEweb-6c8f6b7f84 0 0 0 40mweb-7cc58d88d9 3 3 3 74m

版权声明：本文内容由网络用户投稿，版权归原作者所有，本站不拥有其著作权，亦不承担相应法律责任。如果您发现本站中有涉嫌抄袭或描述失实的内容，请联系我们jiasou666@gmail.com 处理，核实后本网站将在24小时内删除侵权内容。