EKS Quickstart App Dev

This repo contains an initial set of cluster components to be installed and configured by eksctl through GitOps.

Components

• AWS load balancer controller -- to easily expose services to the World.
• Cluster autoscaler -- to automatically add/remove nodes to/from your cluster based on its usage.
  • The autoscaler is configured as recommended in the AWS autoscaler docs
  • See also the autoscaler docs for the AWS provider
• Prometheus (its Alertmanager, its operator, its node-exporter, kube-state-metrics, and metrics-server) -- for powerful metrics & alerts.
• Grafana -- for a rich way to visualize metrics via dashboards you can create, explore, and share.
• Kubernetes dashboard -- Kubernetes' standard dashboard.
• Fluentd & Amazon's CloudWatch agent -- for cluster & containers' log collection, aggregation & analytics in CloudWatch.
• podinfo -- a toy demo application.

Pre-requisites

A running EKS cluster with IAM policies for:

• ALB ingress
• auto-scaler
• CloudWatch

These policies can be added to a nodegroup by including the following iam options in your nodegroup config:

nodeGroups:
  - iam:
      withAddonPolicies:
        albIngress: true
        autoScaler: true
        cloudWatch: true

N.B.: policies are configured at the nodegroup level. Therefore, depending on your use case, you may want to:

• add these policies to all nodegroups,
• add node selectors to the ALB ingress, auto-scaler and CloudWatch pods, so that they are deployed on the nodes configured with these policies.

How to access workloads

For security reasons, this quickstart profile does not expose any workload publicly. However, should you want to access one of the workloads, various solutions are possible.

Port-forwarding

You could port-forward into a pod, so that you (and only you) could access it locally.

For example, for demo/podinfo:

• run:

  kubectl --namespace demo port-forward service/podinfo 9898:9898

• go to http://localhost:9898

Ingress

You could expose a service publicly, at your own risks, via ALB ingress.

N.B.: the ALB ingress controller requires services:

• to be of NodePort type,
• to have the following annotations:

  annotations:
    kubernetes.io/ingress.class: alb
    alb.ingress.kubernetes.io/scheme: internet-facing

NodePort services

For any NodePort service:

apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  name: ${name}
  namespace: ${namespace}
  annotations:
    kubernetes.io/ingress.class: alb
    alb.ingress.kubernetes.io/scheme: internet-facing
  labels:
    app: ${service-app-selector}
spec:
  rules:
    - http:
        paths:
          - path: /*
            backend:
              serviceName: ${service-name}
              servicePort: 80

A few minutes after deploying the above Ingress object, you should be able to see the public URL for the service:

$ kubectl get ingress --namespace demo podinfo
NAME      HOSTS   ADDRESS                                                                     PORTS   AGE
podinfo   *       xxxxxxxx-${namespace}-${name}-xxxx-xxxxxxxxxx.${region}.elb.amazonaws.com   80      1s

HelmRelease objects

For HelmRelease objects, you would have to configure spec.values.service and spec.values.ingress, e.g. for demo/podinfo:

apiVersion: helm.fluxcd.io/v1
kind: HelmRelease
metadata:
  name: podinfo
  namespace: demo
spec:
  releaseName: podinfo
  chart:
    git: https://github.com/stefanprodan/podinfo
    ref: 3.0.0
    path: charts/podinfo
  values:
    service:
      enabled: true
      type: NodePort
    ingress:
      enabled: true
      annotations:
        kubernetes.io/ingress.class: alb
        alb.ingress.kubernetes.io/scheme: internet-facing
      path: /*

N.B.: the above HelmRelease

• changes the type of podinfo's service from its default value, ClusterIP, to NodePort,
• adds the annotations required for the ALB ingress controller to expose the service, and
• exposes all of podinfo's URLs, so that all assets can be served over HTTP.

A few minutes after deploying the above HelmRelease object, you should be able to see the following Ingress object, and the public URL for podinfo:

$ kubectl get ingress --namespace demo podinfo
NAME      HOSTS   ADDRESS                                                             PORTS   AGE
podinfo   *       xxxxxxxx-demo-podinfo-xxxx-xxxxxxxxxx.${region}.elb.amazonaws.com   80      1s

Securing your endpoints

For a production-grade deployment, it's recommended to secure your endpoints with SSL. See Ingress annotations for SSL.

Any sensitive service that needs to be exposed must have some form of authentication. To add authentication to Grafana for e.g., see Grafana configuration. To add authentication to other components, please consult their documentation.

Get in touch

Create an issue, or login to Weave Community Slack (#eksctl) (signup).

Weaveworks follows the CNCF Code of Conduct. Instances of abusive, harassing, or otherwise unacceptable behavior may be reported by contacting a Weaveworks project maintainer, or Alexis Richardson (alexis@weave.works).