setup-gke-cluster

GKE Cluster Setup for Datadog

Configure a GKE Standard cluster for Datadog monitoring, including egress configuration for private clusters.

Prerequisites

• A running GKE Standard cluster
• gcloud CLI authenticated
• Cluster admin permissions

Instructions

1. Verify cluster access

gcloud container clusters get-credentials <CLUSTER_NAME> --region <REGION>
kubectl get nodes

2. Configure Cloud NAT (required for private clusters)

Private GKE clusters need Cloud NAT for the Datadog Agent to reach intake endpoints (*.datadoghq.com):

# Create a Cloud Router
gcloud compute routers create <ROUTER_NAME> \
  --network=<VPC_NAME> \
  --region=<REGION>

# Create Cloud NAT
gcloud compute routers nats create <NAT_NAME> \
  --router=<ROUTER_NAME> \
  --region=<REGION> \
  --auto-allocate-nat-external-ips \
  --nat-all-subnet-ip-ranges

3. Note your cluster name

gcloud container clusters list --format="value(name)"

You will need this for the install-dd-operator skill to set clusterName in the agent configuration.

Validation

# Verify egress works (from a test pod)
kubectl run test-egress --image=curlimages/curl --rm -it --restart=Never -- \
  curl -s -o /dev/null -w "%{http_code}" https://api.datadoghq.com
# Expected: 200 or 403 (reachable)

Troubleshooting

Pods cannot reach external endpoints

Cause: No Cloud NAT configured for the VPC. Fix: Create Cloud NAT as described in Step 2.