Kind Cluster Setup

This is a simple Kind cluster setup for local development. It is based on the official documentation.

Prerequisites

You should have installed kubectl, Docker, and golang 1.20.

Without installing Go

Go is only used to install kind in a platform-agnostic way. If you would like to skip this prerequisite, you can install kind manually by following the instructions here.

If you manually install kind you still must run the 1.install-kind.sh script to create the cluster.

Scenario

Imagine you are working an organization with multiple teams. Each team uses its own language. We'd like to have a consistent way to add services to our products, regardless of the language used. We'd also like to have a consistent way to deploy our services across teams.

What we'll build

In the ./apps folder you will find three different services, written in three different languages. Each service has its own Dockerfile to build the image.

Cluster topology

We have a 3 node cluster. 1 node is dedicated to running the control plane, and we'll have worker nodes to run our apps.

+---------------------+
| Kubernetes Cluster  |
+---------------------+
|                     |
| +-----------------+ |
| | Control Plane   | |
| | Node            | |
| +-----------------+ |
|                     |
| +-----------------+ |       +-----------------+
| | Worker Node 1   | | <---- | Deployments     |
| +-----------------+ |       +-----------------+
|                     |       
| +-----------------+ |       +-----------------+
| | Worker Node 2   | | <---- | Deployments     |
| +-----------------+ |       +-----------------+
|                     |
+---------------------+

Application topology

We'll create three separate apps and create ingress, load balancing, and horizontal scaling. The next diagram depicts the architecture of the system we'll build.

Keep in mind

• That the apps are running in the same namespace, but they are completely isolated from each other.
• Even though the Pods are part of one deployment. The containers within the pod could be running on either worker node

                             +-----------------+
                             |                 |
                             | Ingress         |
                             | Controller      |
                             |                 |
                             +--------+--------+
                                      |
         +----------------------------+-----------------------+------POTENTIALLY SERVICE 4....N --->
         |                            |                       |
+--------v--------+         +---------v-------+      +-----------------+
|                 |         |                 |      |                 |
| Service Ingress |         | Service Ingress |      | Service Ingress |
| "weather"       |         | "dice"          |      | "pod-metadata"  |
|                 |         |                 |      |                 |
+--------+--------+         +---------+-------+      +--------+--------+
         |                            |                       |
         |                            |                       |
+--------v--------+         +---------v-------+      +--------v--------+
|                 |         |                 |      |                 |
| Deployment      |         | Deployment      |      | Deployment      |
| "weather-app"   |         | "dice-app"      |      | "pod-metadata"  |
|                 |         |                 |      |                 |
| +-------------+ |         | +-------------+ |      | +-------------+ |
| | Pod 1       | |         | | Pod 1       | |      | | Pod 1       | |
| | Container 1 | |         | | Container 1 | |      | | Container 1 | |
| +-------------+ |         | +-------------+ |      | +-------------+ |
| +-------------+ |         | +-------------+ |      | +-------------+ |
| | Pod 2       | |         | | Pod 2       | |      | | Pod 2       | |
| | Container 1 | |         | | Container 1 | |      | | Container 1 | |
| +-------------+ |         | +-------------+ |      | +-------------+ |
| +-------------+ |         +-----------------+      +-----------------+
| +-------------+ |         
| | Pod 3       | |         
| | Container 1 | |         
| +-------------+ |         
+-----------------+         

Setup

To set up the example run the bash commands in ./bin in the specified order.

# This script sets up the kind cluster
bash ./bin/1.install-kind.sh
# This script installs helm
bash ./bin/2.install-helm.sh
# This script installs ingress-nginx
bash ./bin/3.install-ingress-nginx.sh
# This script builds and pushes the images to the registry
# You'd repeat this step and the following step
# each time you make a change to the code
bash ./bin/4.build-push.sh
# This script deploys the apps in the cluster
# you run this script when you want to roll
# out changes.
bash ./bin/5.deploy-k8s.yaml

Deploying the apps

Steps 4 and 5 represent the steps of deploying containers into the cluster. You can repeat these steps each time you make a change to the code. The workflow is as follows:

+--------------+       +------------------+       +------------------+       +------------------+
|              |       |                  |       |                  |       |                  |
|   Coding     | ----> |  Build Image     | ----> |  Push to Registry|       |  Kubernetes Pod  |
|              |       |                  |       |                  |       |                  |
+--------------+       +------------------+       +------------------+       +------------------+
                                                       |                              ^
                                                       |                              |
                                                       +------------------------------+
                                                      Pull Image & Update Pod

Testing the apps

To test the apps you can use the following commands:

# Note the hostnames in the ingress.yaml file weather.test and dice.test respectively
# Add them to your hosts file in /etc/hosts
echo "127.0.0.1 weather.test" | sudo tee -a /etc/hosts > /dev/null
echo "127.0.0.1 dice.test" | sudo tee -a /etc/hosts > /dev/null

curl dice.test:8080/roll-dice
# Output {"roll":4} - you could also visit the url in your browser
curl http://weather.test:8080/WeatherForecast
# [{"date":"2023-11-22T00:31:38.874849+00:00","temperatureC":-13,"temperatureF":9,"summary":"Balmy"},{"date":"2023-11-23T00:31:38.874939+00:00","temperatureC":11,"temperatureF":51,"summary":"Bracing"},{"date":"2023-11-24T00:31:38.8749399+00:00","temperatureC":25,"temperatureF":76,"summary":"Mild"},{"date":"2023-11-25T00:31:38.8749401+00:00","temperatureC":31,"temperatureF":87,"summary":"Freezing"},{"date":"2023-11-26T00:31:38.8749403+00:00","temperatureC":32,"temperatureF":89,"summary":"Scorching"}]

Inspecting the resources in the namespace

When working with Kubernetes, kubectl is an essential tool for managing and inspecting resources. Here are some common kubectl commands you might use to inspect what we've deployed in the namespace: your-app.

Get Information About the Namespace

To get basic information about your namespace:

kubectl get namespace your-app

List All Resources in the Namespace

To list all resources in your-app:

kubectl get all --namespace your-app

View Pods in the Namespace

To list all pods in your-app:

kubectl get pods --namespace your-app

Accessing Logs of a Pod

To access logs of a specific pod (replace pod-name with the actual pod name):

kubectl logs pod-name --namespace your-app

Executing Commands in a Pod

To execute a command in a pod (replace pod-name with the actual pod name and command with your command):

kubectl exec -it pod-name --namespace your-app -- command