kubernetes-deployment

Context

Deploying {{project_name}} ({{project_type}}) to Kubernetes. You will follow a strict 6-phase Deployment Lifecycle Contract.

Instructions

Execute the following phases in order:

Phase 1: Authentication

1. Ensure the KUBECONFIG environment variable is set to a valid configuration file, or the configuration is located in ~/.kube/config.
2. Authenticate to the cluster (if using cloud providers like AWS EKS or GCP GKE, use their respective commands: aws eks update-kubeconfig or gcloud container clusters get-credentials).
3. Verify access by running kubectl cluster-info and kubectl get nodes.

Phase 2: Build

1. Prepare the artifacts for deployment.
2. Build your Docker image: docker build -t <registry>/<image-name>:<tag> .
3. Push the image to your container registry (DockerHub, GHCR, ECR, GCR): docker push <registry>/<image-name>:<tag>

Phase 3: Install / Provisioning

1. Ensure the target Kubernetes namespace exists: kubectl create namespace <namespace> (if it doesn't exist).
2. Create or update necessary ConfigMaps and Secrets: kubectl apply -f k8s/configmap.yaml or use kubectl create secret generic.
3. If using Helm, ensure the Helm chart dependencies are updated: helm dependency update <chart-dir>.

Phase 4: Deploy

1. Ship the artifact to Kubernetes.
2. Using raw manifests: Update the image tag in your Deployment YAML, then run kubectl apply -f k8s/.
3. Using Kustomize: kustomize build k8s/ | kubectl apply -f -.
4. Using Helm: helm upgrade --install <release-name> <chart-dir> --namespace <namespace> --set image.tag=<tag>.

Phase 5: Checking

1. Verify the deployment was successful.
2. Check the rollout status: kubectl rollout status deployment/<deployment-name> -n <namespace>. This command will block until the deployment is successful or fails.
3. Check pod status if there are issues: kubectl get pods -n <namespace>.
4. Inspect pod logs if they are crashing: kubectl logs -l app=<app-label> -n <namespace>.
5. Curl the service or ingress URL if available externally.

Phase 6: Update / Rollback

1. If Phase 5 fails (e.g., rollout gets stuck or pods crashloop), immediately initiate a rollback.
2. Using kubectl: kubectl rollout undo deployment/<deployment-name> -n <namespace>.
3. Using Helm: helm rollback <release-name> -n <namespace>.
4. Note the failure in the progress log.

Validation

• Cluster authentication (kubectl cluster-info) succeeds.
• Docker image builds and pushes successfully.
• Kubernetes rollout status is successful.
• Pods are running and healthy.