Kubernetes Deployment with Helm

Objective

Install, upgrade, and roll back an application on a local Kubernetes cluster using Helm charts and custom values files. By the end of this lab you will understand how Helm packages Kubernetes manifests into versioned releases and how to manage their full lifecycle.

Prerequisites

• Docker installed and running
• kind installed (or minikube as an alternative)
• kubectl installed and on your PATH
• Helm v3 installed (helm version should print v3.x)
• At least 4 GB of free RAM for the local cluster

Architecture

The lab runs a single-node Kubernetes cluster inside a Docker container via kind. Helm talks to the Kubernetes API server, renders the chart templates with your values, and applies the resulting manifests. The chart deploys an NGINX Deployment, a Service, and a ConfigMap. Release metadata is stored as Secrets in the cluster namespace, which is what enables rollbacks.

+--------------------------------------------------+
| Your laptop                                      |
|                                                  |
|  helm CLI ----> Kubernetes API (kind container)  |
|                       |                          |
|                       v                          |
|   +-------------- namespace: web -------------+ |
|   | Deployment (nginx) --> Pods x2            | |
|   | Service (ClusterIP)                       | |
|   | Secret (helm release metadata)            | |
|   +--------------------------------------------+ |
+--------------------------------------------------+

Steps

1. Create a local Kubernetes cluster

kind create cluster --name helm-lab
kubectl cluster-info --context kind-helm-lab
kubectl get nodes

2. Add the Bitnami chart repository

helm repo add bitnami https://charts.bitnami.com/bitnami
helm repo update
helm search repo bitnami/nginx --versions | head -5

3. Inspect the chart before installing

helm show values bitnami/nginx | head -40
helm template my-nginx bitnami/nginx --namespace web | head -60

4. Install the chart with custom values

Create a values override file:

# values-lab.yaml
replicaCount: 2
service:
  type: ClusterIP
resources:
  requests:
    cpu: 100m
    memory: 128Mi
  limits:
    cpu: 250m
    memory: 256Mi

Install the release:

kubectl create namespace web
helm install my-nginx bitnami/nginx \
  --namespace web \
  --values values-lab.yaml \
  --wait

5. Verify the release

helm list --namespace web
kubectl get pods,svc --namespace web
kubectl port-forward svc/my-nginx 8080:80 --namespace web &
curl -s http://localhost:8080 | head -5
kill %1

6. Upgrade the release

Bump the replica count and upgrade:

helm upgrade my-nginx bitnami/nginx \
  --namespace web \
  --values values-lab.yaml \
  --set replicaCount=3 \
  --wait
helm history my-nginx --namespace web

7. Roll back to revision 1

helm rollback my-nginx 1 --namespace web --wait
kubectl get pods --namespace web
helm history my-nginx --namespace web

Expected output

$ helm list --namespace web
NAME      NAMESPACE  REVISION  STATUS    CHART          APP VERSION
my-nginx  web        1         deployed  nginx-18.x.x   1.27.x

$ kubectl get pods --namespace web
NAME                        READY   STATUS    RESTARTS   AGE
my-nginx-7d4b9c8f6d-2xkqp   1/1     Running   0          45s
my-nginx-7d4b9c8f6d-9wlmn   1/1     Running   0          45s

$ helm history my-nginx --namespace web
REVISION  STATUS      DESCRIPTION
1         superseded  Install complete
2         superseded  Upgrade complete
3         deployed    Rollback to 1

Troubleshooting

• Pods stuck in ImagePullBackOff: your machine cannot reach Docker Hub. Check connectivity with docker pull bitnami/nginx:latest, or pre-load the image into kind with kind load docker-image bitnami/nginx:latest --name helm-lab.
• Error: Kubernetes cluster unreachable: your kubeconfig context is wrong. Run kubectl config use-context kind-helm-lab and retry.
• helm install hangs with --wait: pods are not becoming Ready. Inspect with kubectl describe pod -n web and look at Events — usually insufficient memory on the kind node.
• port-forward connection refused: the Service has no ready endpoints yet. Confirm with kubectl get endpoints my-nginx -n web before curling.
• Chart version not found: stale repo cache. Run helm repo update and search again.

Cleanup

helm uninstall my-nginx --namespace web
kubectl delete namespace web
kind delete cluster --name helm-lab
rm -f values-lab.yaml