Typical Deployments for cluster validation ¶

Commands below are written for Helm v3. Same commands cannot be used for Helm v2. Please adapt the commands as needed.

Ingress Controller - Ingress controller should be installed first so that other components can be accessed via ingress.

helm repo add ingress-nginx https://kubernetes.github.io/ingress-nginx
helm repo update

kubectl create ns ingress-nginx
# if deploying on AWS KOPS. This setups HTTPS and ACM certificate. (Must adjust ACM cert ARN in the yaml file)
helm upgrade --install ingress-nginx ingress-nginx/ingress-nginx -n ingress-nginx -f deploy/ingress.yaml
# if deploying on Azure/GCP
helm upgrade --install ingress-nginx ingress-nginx/ingress-nginx -n ingress-nginx -f deploy/ingress-ctrl-non-aws.yaml

Kubernetes Dashboard - Kubernete Dashboard.

cd deploy
helm repo add kubernetes-dashboard https://kubernetes.github.io/dashboard/
helm repo update
kubectl create ns dashboard
helm upgrade --install k8s-dashboard kubernetes-dashboard/kubernetes-dashboard -f k8s-dashboard.yaml -n dashboard

Monitoring

cd deploy
kubectl create ns monitoring

# WARN: In some kubernetes distributions, you may need to add a default StorageClass. If prometheus pods in below install remain pending for PVCs.. check for storageclass first.
# use storageclass-*.yaml to create the same first
# install prometheus
helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
helm repo update
helm upgrade --install prometheus prometheus-community/prometheus --version 13.0.1 -f prometheus.yaml -n monitoring

# install grafana
helm repo add grafana https://grafana.github.io/helm-charts
helm repo update
helm upgrade --install grafana grafana/grafana --version 6.1.16 -f grafana.yaml -n monitoring

Connect Grafana with Prometheus and start monitoring your cluster

After logging into grafana as admin, you can add datasource of type prometheus. For the server URL, give value as http://prometheus-server/prometheus and save.
Once prometheus is connected, you can look at monitoring data by importing some of the popular grafana dashboards like 9614 (Node Exporter Full), 315 (kubernetes cluster monitoring), 10518, 3149, 4686, App metrics 1471, 8685, 5228, etc. There are way too many dashboards to list all good ones. Please explore and learn.
podinfo deployment

cd deploy
kubectl apply -k github.com/stefanprodan/podinfo//kustomize
kubectl apply -f podinfo.yaml

Autoscaling testing (both HPA and cluster) For HPA installation and testing please go to HPA page. Cluster Autoscaler FIXME
Storage testing Longhorn FIXME