nsudhanva / node-js-kubernetes-helm

Scalable Node.js with Kubernetes, Helm, Grafana, Prometheus, Jaegar, Zipkin and OpenTracing

Scalable Node.js with Kubernetes, Helm, Grafana, Prometheus, Jaegar, Zipkin and OpenTracing

Requirements

• Kubernetes (or minikube)
• kubectl
• Docker

Helpers

• Re-using local Docker daemon with minikube: eval $(minikube docker-env) (run it once before Docker build)
• On OSX: To base64: pbpaste | base64 | pbcopy and From base64: pbcopy | base64 --decode
• minikube start and minikube stop

Tasks

1. Build Docker image

Re-using local Docker daemon with minikube:

eval $(minikube docker-env)

Building Docker image:

cd src/gateway
docker build -t my-co/gateway:v1 .

2. Create Secret, Deployment and Service in Kubernetes

kubectl create -f k8s
kubectl get deployment
kubectl get deployment <deployment-name> -o yaml
kubectl edit deployment <deployment-name>
kubectl get secret
kubectl get secret <secret-name> -o yaml
kubectl get service
kubectl get service <service-name> -o yaml

3. Get running pods

kubectl get pod
kubectl describe pod <pod-name>

Edit replica number in gateway Deployment and get pods again.
WARNING: Always edit yaml files in Kubernetes via kubectl, it's not synchronized with your local k8s folder.

4. Kill one pod

kubectl get pod
kubectl delete pod <pod-name>
kubectl get pod

5. Test networking

Go to inside a running pod:

kubectl get pod
kubectl exec <pod-name> -it -- sh
nslookup kubernetes gateway
curl <host>:3001
curl <ip>:3001
env | grep GATEWAY

From minikibe:

minikube service gateway

6. Set Horizontal Pod Autoscaler (HPA)

Do you have a running Heapster (collects metrics for autoscaler)?

kubectl get pod -n kube-system

If not, enable it:

$ minikube addons list
- addon-manager: enabled
- dashboard: enabled
- kube-dns: enabled
- heapster: disabled
- registry-creds: disabled

# minikube must be running for these commands to take effect
$ minikube addons enable heapster
heapster was successfully enabled

# this may take some time
$ kubectl get pod -n kube-system
NAME                          READY     STATUS              RESTARTS   AGE
heapster-2b7mt                1/1       Running             0          4m

kubectl autoscale deployment gateway --cpu-percent=50 --min=1 --max=10
kubectl get hpa
kubectl get hpa gateway -o yaml
ab -n 10000 -c 100 <IP>

HPA runs every 2 minutes by default.

7. Templating

Checkout Helm and anchor.

Install helm, then:

$ helm init
Creating /Users/foo/.helm
Creating /Users/foo/.helm/repository
Creating /Users/foo/.helm/repository/cache
Creating /Users/foo/.helm/repository/local
Creating /Users/foo/.helm/plugins
Creating /Users/foo/.helm/starters
Creating /Users/foo/.helm/cache/archive
Creating /Users/foo/.helm/repository/repositories.yaml
$HELM_HOME has been configured at /Users/foo/.helm.

# this will take a while
$ kubectl get pod -n kube-system
NAME                             READY     STATUS    RESTARTS   AGE
tiller-deploy-1936853538-3brdf   1/1       Running   0          7m

Backup your infrastructure:

node chart.js

Your templates will be saved in gateway/templates/

helm package gateway

On a fresh cluster run:

helm install gateway

and your infrastructure should be restored.