k8s基础维护命令

基础维护

# 当需要对主机进行维护升级时,首先将节点主机设置成不可调度模式:

kubectl cordon[nodeid]

 

[root@ziji-k8s-master1-192-168-1-16 coredns]# kubectl cordon 192.168.1.17

node/192.168.1.17 cordoned




[root@ziji-k8s-master1-192-168-1-16 coredns]# kubectl  get node

NAME           STATUS                     ROLES         AGE       VERSION
192.168.1.16   Ready,SchedulingDisabled   master        25d       v1.11.4
192.168.1.17   Ready,SchedulingDisabled   node          27d       v1.11.4
192.168.1.18   Ready                      node          27d       v1.11.4
192.168.1.19   Ready                      node          27d       v1.11.4
192.168.1.20   Ready                      dc1-db,node   15d       v1.11.4

 

# 然后需要将主机上正在运行的容器驱赶到其它可用节点:

 kubectl drain [nodeid]

 

# 给予900秒宽限期优雅的调度

kubectl drain node1.k8s.novalocal --grace-period=120

 

# 当容器迁移完毕后,运维人员可以对该主机进行操作,配置升级性能参数调优等等。当对主机的维护操作完毕后, 再将主机设置成可调度模式:

kubectl uncordon [nodeid]

#强制删除已经Terminating   pod

kubectl get  pod  -n zk-kafka
kubectl  delete zookeeper-2-4n426 -n zk-kafka   --grace-period=0 --force

 

删除Evicted (被驱逐的pod)

kubectl get pods -n apache-test | grep Evicted | awk '{print $1}' | xargs kubectl delete pod  -n apache-test

 

 

更新操作

注意:当且仅当部署的pod模板(即.spec.template)更改时,才会触发Deployment的部署,例如,如果更新模板的标签或容器图像。其他更新(例如扩展部署)不会触发部署。

kubectl set image deployment/web-service web-service=web-service:v1.9.1 --record
deployment "web-service" image updated

 

也可以进行编辑进行更改

$ kubectl edit deployment/web-service  -n  ziji
deployment "nginx-deployment" edited

 

回滚操作< deployment|daemonset >

查看版本发布历史

kubectl rollout history deployment/web-service -n ziji
deployments "web-service"

REVISION  CHANGE-CAUSE
23        <none>
27        <none>
28        <none>

 

回滚到上一个版本

kubectl rollout undo deployment  -n ziji web-service

 

#查看deployment对象信息

kubectl describe deployment -n ziji web-service

Name:                   web-service
Namespace:              ziji
CreationTimestamp:      Sun, 02 Dec 2018 18:31:43 +0800
Labels:                 app=apache-v1
                        name=web-service
Annotations:            deployment.kubernetes.io/revision=28

 

#查看replicasets对象信息

kubectl get rs -o wide -n ziji

 

回滚到指定版本

kubectl rollout undo deployment/web-service   --to-revision=27

 

查看发布的状态

kubectl  rollout  status   deployment/web-service -n ziji
deployment "web-service" successfully rolled out

 

扩容操作

设置自动扩容(水平拉伸), 当pod cpu使用率超过10%, 最小创建pod数为1, 最大3

kubectl  autoscale deployment -n ziji  web-service --min=1 --max=3 --cpu-percent=10

 

指定扩容pod副本数量

kubectl  scale  deployment -n ziji  web-service  --replicas  3

滚动更新

spec:
  replicas: 10
  selector:
    matchLabels:
      service: http-server

  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 1
  minReadySeconds: 5

 

暂停滚动更新

kubectl rollout pause deployment <deployment>

 

继续滚动更新

kubectl rollout resume deployment <deployment>

 

保留版本次数

...

spec:
  replicas: 10
  selector:
    matchLabels:
      service: http-server

  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 1
  minReadySeconds: 5
  revisionHistoryLimit: 10

...

 

常用命令

  • 创建:kubectl run <name> –image=<image> 或者 kubectl create -f manifest.yaml
  • 查询:kubectl get <resource>
  • 更新 kubectl set 或者 kubectl patch
  • 删除:kubectl delete <resource> <name> 或者 kubectl delete -f manifest.yaml
  • 查询 Pod IP:kubectl get pod <pod-name> -o jsonpath='{.status.podIP}’
  • 容器内执行命令:kubectl exec -ti <pod-name> sh
  • 容器日志:kubectl logs [-f] <pod-name>
  • 导出服务:kubectl expose deploy <name> –port=80
  • Base64 解码:
kubectl get secret SECRET -o go-template='{{ .data.KEY | base64decode }}'

 

查看endpoints

[root@feiba-k8s-master1-192-168-1-16 ~]# kubectl get ep -n ziji-test

NAME                                    ENDPOINTS                                                  AGE
glusterfs-dynamic-pvc-zup-config        192.168.1.12:1,192.168.1.13:1,192.168.1.14:1 + 1 more...   34d
glusterfs-dynamic-pvc-zup-gateway       192.168.1.12:1,192.168.1.13:1,192.168.1.14:1 + 1 more...   34d
glusterfs-dynamic-pvc-zup-payment-web   192.168.1.12:1,192.168.1.13:1,192.168.1.14:1 + 1 more...   32d
glusterfs-dynamic-pvc-zup-task          192.168.1.12:1,192.168.1.13:1,192.168.1.14:1 + 1 more...   33d
games                                   10.96.20.3:8099                                            4d
gateway                                 10.96.20.10:6674                                           18d
gateway-admin                           10.96.66.7:5674                                            4d
merchant                                10.96.20.8:8080                                            33d
mining                                  10.96.66.5:8085                                            33d
payment-web                             10.96.20.13:8084                                           32d
register                                10.96.20.7:8888                                            18d
task                                    10.96.20.6:8083                                            33d

 

[root@ziji-k8s-master1-192-168-1-16 ~]# kubectl get endpoints  mining  -n test

NAME         ENDPOINTS         AGE
mining   10.96.66.5:8085   33d

 

日志查看

kubectl logs nginx
kubectl logs -f -c ruby web-1

 

连接到一个正在运行的容器

kubectl attach 用于连接到一个正在运行的容器

 

文件拷贝

kubectl cp /tmp/foo_dir <some-pod>:/tmp/bar_dir     #拷贝文件到容器中
kubectl cp <some-namespace>/<some-pod>:/tmp/foo /tmp/bar  #从容器拷贝文件到本地

 

反亲和性

      affinity:
        podAntiAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
          - labelSelector:
              matchExpressions:
              - key: app
                operator: In
                values:
                - <APP_NAME>
            topologyKey: "kubernetes.io/hostname"

 

发表评论

电子邮件地址不会被公开。 必填项已用*标注

昵称 *