Skip to content

Latest commit

 

History

History

eks-clusters

README.md

Prepare EKS clusters

Prerequisite

Create Locust cluster

1. Settings

# Set optional environment variables
export AWS_PROFILE="YOUR_PROFILE" # If not, use 'default' profile
export AWS_REGION="YOUR_REGION"   # ex. ap-northeast-2

# Set common environment variables
export TARGET_GROUP_NAME="locust"
export TARGET_CLUSTER_NAME="awsblog-loadtest-${TARGET_GROUP_NAME}"
export TARGET_REGION="${AWS_REGION}"

# Set specific environment variables
export TARGET_EKS_VERSION="1.21"

# Check
cat <<EOF
_______________________________________________
* AWS_PROFILE : ${AWS_PROFILE:-(default)}
* AWS_REGION  : ${AWS_REGION:-(invalid!)}
_______________________________________________
* TARGET_GROUP_NAME   : ${TARGET_GROUP_NAME}
* TARGET_CLUSTER_NAME : ${TARGET_CLUSTER_NAME}
* TARGET_REGION       : ${TARGET_REGION:-(invalid!)}
* TARGET_EKS_VERSION  : ${TARGET_EKS_VERSION}
EOF

Notice
If you use the variable (TARGET_EKS_VERSION) about the Kubernetes version,
maybe you need to check the version of kubectl and eksctl.
We recommand to update the latest for matching those versions.

2. Create EKS cluster Configuration file

# Move to 'eks-clusters' directory from root of repository
cd groundwork/eks-clusters

# Create ClusterConfig file
cat eks-cluster-template.yaml | envsubst > "${TARGET_CLUSTER_NAME}.yaml"

3. Provision Locust cluster with eksctl

# Create EKS cluster
eksctl create cluster --with-oidc -f "${TARGET_CLUSTER_NAME}.yaml"

Notice.
This step takes about 20 minutes.
Because it include that create all things (from VPC to EKS cluster).

4. Check the provision with connection

# Create kubeconfig for your kubectl
eksctl utils write-kubeconfig \
  --cluster "${TARGET_CLUSTER_NAME}" \
  --region "${TARGET_REGION}" \
  --profile "${AWS_PROFILE:-(default)}"
# Check kubeconfig context
kubectl config current-context

# Like this..
# <IAM_ROLE>@<TARGET_CLUSTER_NAME>.<TARGET_REGION>.eksctl.io
# Check current nodes
kubectl get nodes -L position --sort-by=.metadata.labels.position

# Like this..
# NAME                                                 STATUS   ROLES    AGE     VERSION                POSITION
# ip-xxx-xxx-xxx-xxx.TARGET_REGION.compute.internal    Ready    <none>   46s     MATCH_ON_EKS_VERSION   addon
# ip-xxx-xxx-xxx-xxx.TARGET_REGION.compute.internal    Ready    <none>   48s     MATCH_ON_EKS_VERSION   locust
# ip-xxx-xxx-xxx-xxx.TARGET_REGION.compute.internal    Ready    <none>   52s     MATCH_ON_EKS_VERSION   locust
# ip-xxx-xxx-xxx-xxx.TARGET_REGION.compute.internal    Ready    <none>   44s     MATCH_ON_EKS_VERSION   locust

5. Set IdP federation for OIDC provider associated with cluster

eksctl utils associate-iam-oidc-provider --cluster=${TARGET_CLUSTER_NAME} --approve

(option) Add identity to auth ConfigMap on other IAM user

# Set target ARN
export IAM_USER_ARN="USER_OR_ROLE_ARN" # e.g arn:aws:iam::111122223333:user/USER_NAME_A
export MAP_USERNAME="admin-$(echo ${IAM_USER_ARN} | awk -F"/" '{print $2}')"

# Check
cat <<EOF
_______________________________________________________
* TARGET_CLUSTER_NAME : ${TARGET_CLUSTER_NAME}
* IAM_USER_ARN : ${IAM_USER_ARN}
* MAP_USERNAME : ${MAP_USERNAME}
EOF
# Add identity target AWS IAM ARN to EKS Auth ConfigMap
eksctl create iamidentitymapping \
    --cluster ${TARGET_CLUSTER_NAME} \
    --group system:masters \
    --arn ${IAM_USER_ARN} \
    --username ${MAP_USERNAME}
# Check
kubectl -n kube-system get cm/aws-auth -o json | jq ".data.mapUsers" | jq -r

# Like this..
# - groups:
#   - system:masters
#   userarn: arn:aws:iam::111122223333:user/USER_NAME_A
#   username: admin-USER_NAME_A
# ...

Create workload cluster

This is optional step when you need workload cluster for load testing.

# Set environment variables
export TARGET_GROUP_NAME="workload"
export TARGET_CLUSTER_NAME="awsblog-loadtest-${TARGET_GROUP_NAME}"
export TARGET_REGION="ap-northeast-2"
export TARGET_EKS_VERSION="1.21"

Then follow up (pervious contents from step.2)

Results