有時候我們想更新一些服務器，或者維護/更新一些鏡像的時候，需要暫停一部分node節點，正確操作步驟如下：
1.獲取節點列表
root@k8s-master1:~# kubectl get nodes -o wide
NAME STATUS ROLES AGE VERSION INTERNAL-IP EXTERNAL-IP OS-IMAGE KERNEL-VERSION CONTAINER-RUNTIME
k8s-master1 Ready control-plane,master 103d v1.22.0 192.168.255.100 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
k8s-master2 Ready control-plane,master 103d v1.22.0 192.168.255.102 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
k8s-node1 Ready <none> 103d v1.22.0 192.168.255.103 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
k8s-node2 Ready <none> 103d v1.22.0 192.168.255.104 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
2.設置不可調度
root@k8s-master1:~# kubectl cordon k8s-node2
node/k8s-node2 cordoned
root@k8s-master1:~# kubectl get nodes -o wide
NAME STATUS ROLES AGE VERSION INTERNAL-IP EXTERNAL-IP OS-IMAGE KERNEL-VERSION CONTAINER-RUNTIME
k8s-master1 Ready control-plane,master 103d v1.22.0 192.168.255.100 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
k8s-master2 Ready control-plane,master 103d v1.22.0 192.168.255.102 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
k8s-node1 Ready <none> 103d v1.22.0 192.168.255.103 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
k8s-node2 Ready,SchedulingDisabled <none> 103d v1.22.0 192.168.255.104 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
root@k8s-master1:~#
可以看到k8s-node2節點已經被暫停調度（SchedulingDisabled）
3.驅逐節點上運行的Pod
如果需要驅逐節點上的pod 可執行
root@k8s-master1:~# kubectl drain k8s-node2
node/k8s-node2 already cordoned
DEPRECATED WARNING: Aborting the drain command in a list of nodes will be deprecated in v1.23.
The new behavior will make the drain command go through all nodes even if one or more nodes failed during the drain.
For now, users can try such experience via: --ignore-errors
error: unable to drain node "k8s-node2", aborting command...
There are pending nodes to be drained:
k8s-node2
error: cannot delete DaemonSet-managed Pods (use --ignore-daemonsets to ignore): kube-system/kube-flannel-ds-wdwzs, kube-system/kube-proxy-dk4vn
root@k8s-master1:~#
注意：若node節點上存在daemonsets控制器創建的pod,則需要使用--ignore-daemonsets忽略錯誤錯誤警告
root@k8s-master1:~# kubectl drain k8s-node2 --ignore-daemonsets
node/k8s-node2 already cordoned
WARNING: ignoring DaemonSet-managed Pods: kube-system/kube-flannel-ds-wdwzs, kube-system/kube-proxy-dk4vn
node/k8s-node2 drained
root@k8s-master1:~#
查看pod，可以看到已經沒有pod運行在k8s-node2節點
root@k8s-master1:~# kubectl get pod -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
php-fpm-nginx-5d8b7fd989-59ljf 2/2 Running 0 91m 10.10.3.243 k8s-node1 <none> <none>
php-fpm-nginx-5d8b7fd989-kfmln 2/2 Running 22 (22h ago) 24d 10.10.3.238 k8s-node1 <none> <none>
redis-svc-866fb777f8-4ps2s 1/1 Running 11 (22h ago) 24d 10.10.3.240 k8s-node1 <none> <none>
redis-svc-866fb777f8-lflkv 1/1 Running 11 (22h ago) 24d 10.10.3.239 k8s-node1 <none> <none>
test-www-7fcc4d5595-4mg5j 2/2 Running 0 91m 10.10.3.242 k8s-node1 <none> <none>
test-www-7fcc4d5595-kcllj 2/2 Running 0 91m 10.10.3.241 k8s-node1 <none> <none>
4、恢復可調度
當需要重新調度node節點的時候，執行
root@k8s-master1:~# kubectl uncordon k8s-node2
node/k8s-node2 uncordoned
查看node狀態
root@k8s-master1:~# kubectl get nodes -o wide
NAME STATUS ROLES AGE VERSION INTERNAL-IP EXTERNAL-IP OS-IMAGE KERNEL-VERSION CONTAINER-RUNTIME
k8s-master1 Ready control-plane,master 103d v1.22.0 192.168.255.100 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
k8s-master2 Ready control-plane,master 103d v1.22.0 192.168.255.102 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
k8s-node1 Ready <none> 103d v1.22.0 192.168.255.103 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
k8s-node2 Ready <none> 103d v1.22.0 192.168.255.104 <none> Ubuntu 18.04.5 LTS 4.15.0-154-generic docker://20.10.8
root@k8s-master1:~#
5、刪除節點
如果需要直接刪除節點（此方式比較暴力，一般情況不推薦使用）
kubectl delete nodes k8s-node2
如果刪除的node想重新加入到k8s集群中 則需要master重新生成加入token即可