Create eks cluster and bootstrap gitlab project with flux

bootstrap.sh

#!/bin/env bash

set -e

set -a
source .env
set +a

# Check if the cluster already exists
CLUSTER_STATUS=$(eksctl get cluster --name ${EKS_CLUSTER_NAME} --region ${EKS_CLUSTER_REGION} --output json | jq -r '.[0].Status')

if [ "$CLUSTER_STATUS" == "CREATING" ]; then
  echo "Cluster ${EKS_CLUSTER_NAME} is currently being created. Waiting for it to become active..."
  eksctl utils wait-nodes --cluster ${EKS_CLUSTER_NAME} --nodes-to-wait-for 1
elif [ "$CLUSTER_STATUS" == "DELETING" ]; then
  echo "Cluster ${EKS_CLUSTER_NAME} is currently being deleted. Exiting."
  exit 1
elif [ "$CLUSTER_STATUS" == "ACTIVE" ]; then
  echo "Cluster ${EKS_CLUSTER_NAME} already exists and is active."
else
  echo "Creating cluster ${EKS_CLUSTER_NAME}..."
  eksctl create cluster \
    --name ${EKS_CLUSTER_NAME} \
    --version ${EKS_CLUSTER_VERSION} \
    --region ${EKS_CLUSTER_REGION} \
    --node-type t3.large \
    --nodes 1 \
    --node-volume-size 128 \
    --nodes-min 1 \
    --nodes-max 3
fi

eksctl utils write-kubeconfig --cluster=${EKS_CLUSTER_NAME} --region=${EKS_CLUSTER_REGION}
kubectl cluster-info

TIMEOUT=0
until kubectl get nodes --no-headers | awk '{print $2}' | grep -q 'Ready' || [ $TIMEOUT -eq 60 ]; do
  echo "Waiting for nodes to be ready..."
  sleep 10
  TIMEOUT=$((TIMEOUT + 10))
done

if [ $TIMEOUT -eq 60 ]; then
  echo "Timeout waiting for nodes to be ready"
  exit 1
fi

# Wait for Metrics Server to be ready
wait_for_metrics_server() {
  local -r metrics_server_namespace="kube-system"
  local -r metrics_server_deployment="metrics-server"
  local -r timeout_seconds=300 # 5 minutes

  local start_time=$(date +%s)
  local elapsed_time=0

  local get_deployment_cmd=$(kubectl get deployment -n $metrics_server_namespace $metrics_server_deployment)
  local get_pods_cmd=$(kubectl get pods -n $metrics_server_namespace -l app.kubernetes.io/name=metrics-server -o jsonpath='{.items[*].status.containerStatuses[*].ready}')

  echo "Waiting for Metrics Server to be ready..."

  while true; do
    # Check if the Metrics Server deployment exists
    if [ -z "$get_deployment_cmd" ]; then
      echo "Metrics Server deployment not found."
      return 1
    fi

    # Check if the Metrics Server pods are ready
    if echo "$get_pods_cmd" | tr ' ' '\n' | grep -vq 'false'; then
      echo "Metrics Server is ready!"
      return 0
    fi

    # Check if the timeout has been reached
    elapsed_time=$(($(date +%s) - $start_time))
    if [ "$elapsed_time" -gt "$timeout_seconds" ]; then
      echo "Timeout reached while waiting for Metrics Server to be ready."
      return 1
    fi

    sleep 5
  done
}

# Deploy Metrics Server (replace this with your actual deployment command)
kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/latest/download/components.yaml

# Wait for Metrics Server to be ready
wait_for_metrics_server


echo ${GITLAB_TOKEN} | flux bootstrap gitlab \
  --owner=${GITLAB_GROUP} \
  --repository=${GITLAB_PROJECT} \
  --branch=${GITLAB_BRANCH} \
  --path=./workspace/clusters/management \
  --timeout=15m \
  --interval=20m