go app with helm charts

This repo consists of a simple go web application. It also has a helm chart that deploys it to a kubernetes cluster along with a config map. The objective of the repository is to understand how to use configmaps along with kubernetes apps

Running the helm chart

Pre-requisites

1. Docker
2. minikube
3. kubernetes
4. helm3
5. go

Running the code in kubernetes

1. Start minikube

    minikube start

2. Run the following command to be able to build the docker image in the minikube docker env

    eval $(minikube docker-env)

3. Build the docker image in the same terminal where the above command was executed

    docker build -t gok8 .

4. Install the application using helm The syntax for the following command is helm install <release-name> <chart-location>

    helm install v1 ./go-helm

Verifying the installed code

1. Run kubectl get pods to check the running pods. The output should be something like

    kubectl get pods
    NAME                          READY   STATUS    RESTARTS   AGE
    go-web-app-54b544476d-mgktm   1/1     Running   0          5s
    go-web-app-54b544476d-vcvjj   1/1     Running   0          5s

2. Since the app exposes 2 rest endpoints namely / and /config one can access it by running

    minikube service go-web-service

   where go-web-service is the name of the kubernetes service we deployed The above command establishes a ssh tunnel with our kubernetes app and opens a browser

   The output of the above command is something like

    minikube service go-web-service
    |-----------|----------------|-------------|---------------------------|
    | NAMESPACE |      NAME      | TARGET PORT |            URL            |
    |-----------|----------------|-------------|---------------------------|
    | default   | go-web-service | http/80     | http://192.168.49.2:32029 |
    |-----------|----------------|-------------|---------------------------|
    🏃  Starting tunnel for service go-web-service.
    |-----------|----------------|-------------|------------------------|
    | NAMESPACE |      NAME      | TARGET PORT |          URL           |
    |-----------|----------------|-------------|------------------------|
    | default   | go-web-service |             | http://127.0.0.1:57922 |
    |-----------|----------------|-------------|------------------------|
    🎉  Opening service default/go-web-service in default browser...
    ❗  Because you are using a Docker driver on darwin, the terminal needs to be open to run it.

3. One can now navigate to http://127.0.0.1:57922/config to view the config map that was loaded Note that the port number will be different every time

Changing the config map

The deployment.yaml has an annotation that is tied to the sha256sum of the client_config.json file. This is so that the pods restart automatically if there is any change to the client_config.json file

template:
    metadata:
      labels:
        name: go-web-app
      annotations:
        checksum/config: {{ .Files.Get "client_config.json" | sha256sum }}

So now modify the client_config.json and upgrade the release by running

helm upgrade v1 ./go-helm

Immediately run kubectl get pods and you should see that the old pods are being terminated and new pods get created due to change in config map

kubectl get pods
NAME                          READY   STATUS        RESTARTS   AGE
go-web-app-54b544476d-mgktm   0/1     Terminating   0          12m
go-web-app-54b544476d-vcvjj   0/1     Terminating   0          12m
go-web-app-696994cfc4-76lm5   1/1     Running       0          6s
go-web-app-696994cfc4-pzp97   1/1     Running       0          5s

Verify the updated configs by again navigating to the /config end point of the app

Clean up

To clean up the installation run the following helm command

helm delete <release-name>

In this case it is

helm delete v1