I currently have a Managed Node Group serving my EKS cluster and have added another:
- NodeGroup1 [current - 20gb ec2 disk]
- NodeGroup2 [new - 80gb ec2 disk]
I'd like to migrate my current pods from Node Group 1 to Node Group 2 to allow more disk size for the worker nodes, and then stop using Node Group 1.
I've made the new node group as such:
eksctl create nodegroup --cluster prod --name NodeGroup2 --node-type t3.medium --nodes 2 --nodes-min 0 --nodes-max 4 --enable-ssm --managed --node-volume-size 80
I have a pod disruption budget set up but at the moment I can tolerate downtime. As long as everything makes it over from one managed node group to the new one, I'm happy.
Can I simply do eksctl delete nodegroup NodeGroup1 and it will move everything to the second node group?
Here's what my node groups, deployments and pods look like:
~$ eksctl get nodegroups --cluster prod -o json
[
{
"StackName": "",
"Cluster": "prod",
"Name": "NodeGroup2",
"Status": "ACTIVE",
"MaxSize": 4,
"MinSize": 0,
"DesiredCapacity": 2,
"InstanceType": "t3.medium",
"ImageID": "AL2_x86_64",
"CreationTime": "2021-11-07T04:15:49.595Z",
"NodeInstanceRoleARN": "arn:aws:iam::redacted:role/eks-node-group",
"AutoScalingGroupName": "eks-...1d",
"Version": "1.20"
},
{
"StackName": "",
"Cluster": "prod",
"Name": "NodeGroup1",
"Status": "ACTIVE",
"MaxSize": 4,
"MinSize": 0,
"DesiredCapacity": 2,
"InstanceType": "t3.medium",
"ImageID": "AL2_x86_64",
"CreationTime": "2021-05-25T06:52:25.437Z",
"NodeInstanceRoleARN": "arn:aws:iam::redacted:role/eks-node-group",
"AutoScalingGroupName": "eks-...01",
"Version": "1.20"
}
~$ kubectl get pods -A
NAMESPACE NAME READY STATUS RESTARTS AGE
default grafana-agent 1/1 Running 0 2d23h
default promtail-2a23 1/1 Running 0 3d
default promtail-2vg2 1/1 Running 0 3d
default prod-application-34 1/1 Running 0 3d
default prod-applicationworker-6l 1/1 Running 0 3d
kube-system aws-load-balancer-controller 1/1 Running 0 2d23h
kube-system aws-node-5rzk3 1/1 Running 0 3d
kube-system aws-node-keljg 1/1 Running 0 3d
kube-system cluster-autoscaler 1/1 Running 0 2d23h
kube-system coredns-3254s 1/1 Running 0 3d
kube-system coredns-48grd 1/1 Running 0 2d23h
kube-system kube-proxy-6vx89 1/1 Running 0 3d
kube-system kube-proxy-rqb23 1/1 Running 0 3d
kube-system metrics-server 1/1 Running 0 2d23h
kubernetes-dashboard dashboard-metrics-scraper 1/1 Running 0 2d23h
kubernetes-dashboard kubernetes-dashboard 1/1 Running 0 2d23h
~$ kubectl get deployments -A
NAMESPACE NAME READY UP-TO-DATE AVAILABLE AGE
default grafana-agent 1/1 1 1 5d23h
default prod-application 1/1 1 1 160d
default prod-applicationworker 1/1 1 1 37d
kube-system aws-load-balancer-controller 1/1 1 1 166d
kube-system cluster-autoscaler 1/1 1 1 166d
kube-system coredns 2/2 2 2 167d
kube-system metrics-server 1/1 1 1 166d
kubernetes-dashboard dashboard-metrics-scraper 1/1 1 1 165d
kubernetes-dashboard kubernetes-dashboard 1/1 1 1 165d
If there's a way I can temporarily cordon a whole node group and then drain it, to the other nodegroup (before I delete the first) that would be amazing.
Sorry if this is a simple question but I've ended up reading so many documents that slightly contradict each other.
Cheers, Mike.
