执行kubectl cordon命令时，工作节点上运行的Pod等对象会继续运行，但节点会被标记为SchedulingDisabled状态，此时用户新建的Pod等对象不会调度到此节点。
查看所有节点状态，默认情况下均为Ready状态
kubectl get nodes
NAME STATUS ROLES AGE VERSION
ip-192-168-3-197.cn-northwest-1.compute.internal Ready <none> 10h v1.14.9-eks-1f0ca9
ip-192-168-33-87.cn-northwest-1.compute.internal Ready <none> 10h v1.14.9-eks-1f0ca9
执行kubectl cordon命令使节点进入维护状态，可以看到节点状态变为Ready,SchedulingDisabled
kubectl cordon ip-192-168-3-197.cn-northwest-1.compute.internal
kubectl get nodes
NAME STATUS ROLES AGE VERSION
ip-192-168-3-197.cn-northwest-1.compute.internal Ready,SchedulingDisabled <none> 10h v1.14.9-eks-1f0ca9
ip-192-168-33-87.cn-northwest-1.compute.internal Ready <none> 10h v1.14.9-eks-1f0ca9
创建Nginx Deployment，验证节点是否可以被调度
kubectl create deployment nginx --image=nginx
kubectl get deployment nginx
NAME READY UP-TO-DATE AVAILABLE AGE
nginx 1/1 1 1 9s
kubectl scale deployment nginx --replicas=4
kubectl get deployment nginx
NAME READY UP-TO-DATE AVAILABLE AGE
nginx 4/4 4 4 35s
可以看到，所有Pod均调度到另外1台节点
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nginx-554b9c67f9-b59sc 1/1 Running 0 2s 192.168.81.68 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-kr828 1/1 Running 0 25s 192.168.91.54 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-qjdgp 1/1 Running 0 2s 192.168.87.179 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-w4vg4 1/1 Running 0 2s 192.168.83.202 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
执行kubectl uncordon，将工作节点恢复到节点组中
kubectl uncordon ip-192-168-3-197.cn-northwest-1.compute.internal
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nginx-65f88748fd-fqzm2 1/1 Running 0 7m59s 192.168.24.171 ip-192-168-3-197.cn-northwest-1.compute.internal <none> <none>
nginx-65f88748fd-glm4q 1/1 Running 0 8m39s 192.168.44.164 ip-192-168-33-87.cn-northwest-1.compute.internal <none> <none>
调整Nginx Deployment副本数量，使Pod重新分布
kubectl scale deployment nginx --replicas=1
kubectl scale deployment nginx --replicas=4
再次查看Pod在节点上的分布状态，可以看到Pod被调度到所有节点
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nginx-554b9c67f9-7w24w 1/1 Running 0 3s 192.168.9.108 ip-192-168-30-112.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-dszwj 1/1 Running 0 3s 192.168.28.3 ip-192-168-30-112.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-kr828 1/1 Running 0 5m13s 192.168.91.54 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-n29br 1/1 Running 0 3s 192.168.85.206 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
执行Kubectl drain命令时，工作节点会被标记SchedulingDisabled，用户新建的Pod等对象不可调度到节点，同时节点上运行的Pod等对象也会被强行驱逐。
查看所有节点状态，默认情况下均为Ready状态
kubectl get nodes
NAME STATUS ROLES AGE VERSION
ip-192-168-3-197.cn-northwest-1.compute.internal Ready <none> 10h v1.14.9-eks-1f0ca9
ip-192-168-33-87.cn-northwest-1.compute.internal Ready <none> 10h v1.14.9-eks-1f0ca9
查看Pod在节点上的分布情况
kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nginx-554b9c67f9-7w24w 1/1 Running 0 10m 192.168.9.108 ip-192-168-30-112.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-dszwj 1/1 Running 0 10m 192.168.28.3 ip-192-168-30-112.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-kr828 1/1 Running 0 15m 192.168.91.54 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-n29br 1/1 Running 0 10m 192.168.85.206 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
执行kubectl drain命令，使节点进入维护状态，同时驱逐节点上运行的Pod
注意：如果提示error: cannot delete Pods with local storage，请在命令后加上--delete-local-data参数
kubectl drain ip-192-168-30-112.cn-northwest-1.compute.internal --ignore-daemonsets --delete-local-data
node/ip-192-168-30-112.cn-northwest-1.compute.internal cordoned
WARNING: ignoring DaemonSet-managed Pods: kube-system/aws-node-rdzrz, kube-system/calico-node-m242x, kube-system/ebs-csi-node-fbqnr, kube-system/efs-csi-node-j6q5f, kube-system/kube-proxy-9pz4h
evicting pod "ebs-csi-controller-6dcc4dc6f4-6rb86"
evicting pod "nginx-554b9c67f9-7w24w"
evicting pod "nginx-554b9c67f9-dszwj"
evicting pod "calico-typha-horizontal-autoscaler-5ddb46bbd8-5d9f4"
evicting pod "cluster-autoscaler-989ffb-pvkjw"
evicting pod "coredns-6565755d58-c7q7s"
pod/coredns-6565755d58-c7q7s evicted
pod/calico-typha-horizontal-autoscaler-5ddb46bbd8-5d9f4 evicted
pod/nginx-554b9c67f9-dszwj evicted
pod/ebs-csi-controller-6dcc4dc6f4-6rb86 evicted
pod/nginx-554b9c67f9-7w24w evicted
pod/cluster-autoscaler-989ffb-pvkjw evicted
node/ip-192-168-30-112.cn-northwest-1.compute.internal evicted
查看节点状态
kubectl get nodes
ip-192-168-30-112.cn-northwest-1.compute.internal Ready,SchedulingDisabled <none> 36h v1.15.10-eks-bac369
ip-192-168-85-96.cn-northwest-1.compute.internal Ready <none> 35h v1.15.10-eks-bac369
查看Pod在节点上的分布情况，可以看到Pod已经迁移到另外的节点
kubectl get pods --all-namespaces -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nginx-554b9c67f9-c7r82 0/1 Pending 0 102s <none> <none> <none> <none>
nginx-554b9c67f9-g4d7v 1/1 Running 0 103s 192.168.87.179 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-kr828 1/1 Running 0 20m 192.168.91.54 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-n29br 1/1 Running 0 14m 192.168.85.206 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
将节点恢复到工作节点组中，可以看到，Pod又可以被调度到该节点
kubectl uncordon ip-192-168-3-197.cn-northwest-1.compute.internal
kubectl get pods --all-namespaces -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nginx-554b9c67f9-c7r82 1/1 Running 0 3m49s 192.168.17.18 ip-192-168-30-112.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-g4d7v 1/1 Running 0 3m50s 192.168.87.179 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-kr828 1/1 Running 0 22m 192.168.91.54 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
nginx-554b9c67f9-n29br 1/1 Running 0 17m 192.168.85.206 ip-192-168-85-96.cn-northwest-1.compute.internal <none> <none>
创建1.14版本的集群和托管工作节点组
eksctl create cluster \
--name prod \
--version 1.14 \
--region cn-northwest-1 \
--managed \
--nodegroup-name managed-1-14 \
--node-type t3.medium \
--nodes 2 \
--nodes-min 1 \
--nodes-max 4 \
--alb-ingress-access
参照官方文档，升级EKS集群Kubernetes版本到1.15 https://docs.aws.amazon.com/zh_cn/eks/latest/userguide/update-cluster.html
获取托管节点组名字
eksctl get nodegroups --cluster test
参考输出
CLUSTER NODEGROUP CREATED MIN SIZE MAX SIZE DESIRED CAPACITY INSTANCE TYPE IMAGE ID
test test-nodegroup 2020-04-07T16:16:54Z 2 6 2 t3.micro
查看托管工作节点组AMI发行版本
aws eks describe-nodegroup --cluster-name test --nodegroup-name test-nodegroup | grep releaseVersion
参考输出
"releaseVersion": "1.14.9-20200228",
执行aws eks update-nodegroup-version命令升级AMI发行版本
aws eks update-nodegroup-version --cluster-name test --nodegroup-name test-nodegroup
参考输出
{
"update": {
"id": "ce13106a-5127-33e7-8b6d-07a0b262ebbb",
"status": "InProgress",
"type": "VersionUpdate",
"params": [
{
"type": "Version",
"value": "1.14"
},
{
"type": "ReleaseVersion",
"value": "1.14.9-20200228"
}
],
"createdAt": 1586339867.894,
"errors": []
}
}
大约等待30分钟，升级完成，再次查看托管节点组AMI发行版本
aws eks describe-nodegroup --cluster-name test --nodegroup-name test-nodegroup | grep releaseVersion
参考输出
"releaseVersion": "1.15.10-20200228",
托管节点组升级完成。
假定现有集群中已有名为test的1.14版本的自行管理工作线程节点组，接下来我们将采用" 迁移到新的工作线程节点组 "的方式进行升级。
查看集群中的所有托管节点组和自行管理节点组，
eksctl get nodegroups --cluster test
参考输出如下，NODEGROUP列下名为dev的节点组是提前创建的自行管理节点组
CLUSTER NODEGROUP CREATED MIN SIZE MAX SIZE DESIRED CAPACITY INSTANCE TYPE IMAGE ID
test dev 2020-04-08T11:42:19Z 1 4 2 t3.medium ami-047594a1f70bab3e2
test test-nodegroup 2020-04-07T16:16:54Z 2 6 2 t3.micro
创建新的1.15版本的工作线程节点组
eksctl create nodegroup \
--cluster test \
--version 1.15 \
--name prod-1-15 \
--node-type t3.medium \
--node-ami auto \
--nodes 2 \
--nodes-min 1 \
--nodes-max 4
创建完成后，查看所有节点的状态是否都已经Ready
kubectl get nodes
参考输出
NAME STATUS ROLES AGE VERSION
ip-192-168-16-40.cn-northwest-1.compute.internal Ready <none> 47s v1.15.10-eks-bac369
ip-192-168-17-102.cn-northwest-1.compute.internal Ready <none> 41m v1.14.9-eks-1f0ca9
ip-192-168-55-90.cn-northwest-1.compute.internal Ready <none> 39m v1.15.10-eks-bac369
ip-192-168-64-254.cn-northwest-1.compute.internal Ready <none> 39m v1.15.10-eks-bac369
ip-192-168-75-131.cn-northwest-1.compute.internal Ready <none> 41m v1.14.9-eks-1f0ca9
ip-192-168-91-145.cn-northwest-1.compute.internal Ready <none> 50s v1.15.10-eks-bac369
查看工作线程节点组
eksctl get nodegroups --cluster test
参考输出
CLUSTER NODEGROUP CREATED MIN SIZE MAX SIZE DESIRED CAPACITY INSTANCE TYPE IMAGE ID
test dev 2020-04-08T11:42:19Z 1 4 2 t3.medium ami-047594a1f70bab3e2
test prod-1-15 2020-04-08T12:23:13Z 1 4 2 t3.medium ami-039044adc28bd70e9
test test-nodegroup 2020-04-07T16:16:54Z 2 6 2 t3.micro
使用如下命令，删除旧版本节点组
eksctl delete nodegroup --cluster test --name dev
等待数分钟后，删除成功，再次查看，旧节点组已从列表消失
eksctl get nodegroups --cluster test
CLUSTER NODEGROUP CREATED MIN SIZE MAX SIZE DESIRED CAPACITY INSTANCE TYPE IMAGE ID
test prod-1-15 2020-04-08T12:23:13Z 1 4 2 t3.medium ami-039044adc28bd70e9
test test-nodegroup 2020-04-07T16:16:54Z 2 6 2 t3.micro