Skip to content

Latest commit

 

History

History
108 lines (98 loc) · 4.5 KB

demo.md

File metadata and controls

108 lines (98 loc) · 4.5 KB
Raw

k8s-cluster-autoscaler

GIVEN:

  • A developer desktop with docker & git installed (AWS Cloud9)
  • An EKS cluster named 'cluster-eksctl' has been created via 03/create-cluster-eksctl-existing-vpc-advanced
  • The k8s cluster autoscaler was installed when cluster was created

WHEN:

  • I validate the k8s cluster autoscaler is functioning

THEN:

  • I will create a simple deployment
  • I will validate the NodeGroups ASG min/max settings
  • I will scale the deployment

SO THAT:

  • I can see the cluster add nodes when necessary

Return to Main Readme

REQUIRES

  • 00-setup-cloud9
  • 03/create-cluster-eksctl-existing-vpc-advanced

DEMO

0: Reset Cloud9 Instance environ from previous demo(s).

  • Reset your region & AWS account variables in case you launched a new terminal session:
cd ~/environment/mglab-share-eks/demos/05/k8s-cluster-autoscaler/
export C9_REGION=$(curl --silent http://169.254.169.254/latest/dynamic/instance-identity/document |  grep region | awk -F '"' '{print$4}')
export C9_AWS_ACCT=$(curl -s http://169.254.169.254/latest/dynamic/instance-identity/document | grep accountId | awk -F '"' '{print$4}')
export AWS_ACCESS_KEY_ID=$(cat ~/.aws/credentials | grep aws_access_key_id | awk '{print$3}')
export AWS_SECRET_ACCESS_KEY=$(cat ~/.aws/credentials | grep aws_secret_access_key | awk '{print$3}')
clear
echo $C9_REGION
echo $C9_AWS_ACCT

1: Deploy & Verify the Cluster Autoscaler.

  • Update our kubeconfig to interact with the cluster created in 03/create-cluster-eksctl-existing-vpc-advanced.
eksctl utils write-kubeconfig --cluster cluster-eksctl --region $C9_REGION --authenticator-role-arn arn:aws:iam::${C9_AWS_ACCT}:role/cluster-eksctl-creator-role
kubectl config view --minify | grep 'cluster-name' -A 1
kubectl get ns
  • Confirm the Cluster Autoscaler is functional, use ctrl-c to exit:
kubectl logs deployment.apps/cluster-autoscaler -n kube-system -f
  • Get cluster cluster-autoscaler service account & review IAM Role IRSA Mappings:
kubectl get sa cluster-autoscaler -n kube-system -o yaml
kubectl get sa cluster-autoscaler -n kube-system -o yaml | grep "eks.amazonaws.com/role-arn"
  • Get the Cluster Autoscaler Deployment and review that the 'cluster-autoscaler' K8s service account is being leveraged by the deployment:
kubectl get deployment $(kubectl get deployment -n kube-system | grep cluster-autoscaler | awk '{print$1}') -n kube-system -o yaml
kubectl get deployment $(kubectl get deployment -n kube-system | grep cluster-autoscaler | awk '{print$1}') -n kube-system -o yaml | grep "serviceAccount:"
  • List the EKS Clusters private OIDC provider that is trusted to sign IRSA tokens so the K8s SA can assume the IAM Role:
export OIDC_URL=$(aws eks describe-cluster --name cluster-eksctl --query 'cluster.identity.oidc.issuer' --region $C9_REGION | tr -d '\"')
echo $OIDC_URL
aws iam list-open-id-connect-providers

2: Create & Scale deployment to witness the CA scale EC2 instances up/down.

  • Count Current EC2 Nodes:
kubectl get nodes -o wide | grep -v fargate
  • Create a new K8s deployment scaled to 0:
kubectl create deployment autoscaler-demo --image=public.ecr.aws/u3e9a9s8/nginx:latest
kubectl set resources deploy autoscaler-demo --requests=cpu=200m,memory=200Mi
  • Scale deployment UP:
kubectl scale deployment autoscaler-demo --replicas=50
  • Watch cluster autoscale log events, ctrl-c to exit:
kubectl logs deployment.apps/$(kubectl get deployment -n kube-system | grep cluster-autoscaler | awk '{print$1}') -n kube-system -f
  • Watch pod scale up as nodes are added, ctrl-c to exit:
kubectl get deployment autoscaler-demo --watch
  • Confirm the CA has scaled nodes up:
kubectl get nodes -o wide | grep -v fargate
  • Delete the deployment, it will take considerably longer for the CA to scale nodes down:
kubectl delete deployment autoscaler-demo

DEPENDENTS

CLEANUP

  • Do not cleanup if you plan to run any dependent demos