Project Title

EKS-Metrics-server

Description

Deploy Metrics server on an EKS cluster using Terraform and Helm.

Getting Started

Dependencies

• Docker
• AWS user with programmatic access and high privileges
• Linux terminal
• Deploy an EKS K8 Cluster with Self managed Worker nodes on AWS using Terraform.

Installing

• Clone the repository
• Set environment variable TF_VAR_AWS_PROFILE
• Review terraform variable values in variables.tf, locals.tf
• Override values in the Helm chart through the "chart_values.yaml" file
• Update kubernetes.tf with the AWS S3 bucket name and key name from the output of the EKS K8 Cluster

Executing program

• Configure AWS user with AWS CLI.

docker-compose run --rm aws configure --profile $TF_VAR_AWS_PROFILE

docker-compose run --rm aws sts get-caller-identity

• Specify appropriate Terraform workspace.

docker-compose run --rm terraform workspace show

docker-compose run --rm terraform workspace select default

• Run Terraform apply to create the EKS cluster, k8 worker nodes and related AWS resources.

./run-docker-compose.sh terraform init

./run-docker-compose.sh terraform validate

./run-docker-compose.sh terraform plan

./run-docker-compose.sh terraform apply

• Verify kubectl calls and ensure Metrics server Pod is in Running status.

./run-docker-compose.sh kubectl get pods -A | grep -i metric

• Check out top nodes and pods by CPU & Memory consumption on the EKS cluster.

./run-docker-compose.sh kubectl top node

./run-docker-compose.sh kubectl top pod -A

Help

• metrics-server is unable to authenticate to apiserver

Issue: kubectl top node and pod shows unknown instead of perencentage. The pod log and event has the error: dial tcp ip_address: i/o timeout

Fix:
Check chart_values.yaml

Reference: kubernetes-sigs/metrics-server#278

Authors

Sivanandam Manickavasagam

Version History

• 0.1
  • Initial Release

License

This project is licensed under the MIT License - see the LICENSE file for details

Repo rosters

Stargazers