This repository contains a simple web server implemented in Go, that can retrieve the resource requests and limits of containers in a Kubernetes cluster.
It can be run locally - in that case the authentication is done by accessing the ~/.kube/config file (and taking the credentials from there).
It can be deployed to a Kubernetes cluster as well, in that case it relies on the deployment's service account for authentication.
Before running, first create a build of the application. This can be done with the make build command (or make all to run the unit tests befgore the build).
The resulting binary can be found in the bin/ directory and can be run like: bin/k8s-resource-watcher.
The environment - dev or prod - can be specified using the ENV environmental variable. This will decide which authentication method will be used and on which address the server is listening.
A basic version of the application can be deployed with the kubectl command, following the instructions from the k8s folder.
Using the Helm chart from the helm directory however provides more configuration options and it also creates a service account (which provides the necessary permissions for accessing pod details in the cluster) and a HPA (to scale the deployment if the load increases).
To deploy the chart, run the following command from the helm directory: helm upgrade --install k8s-resource-watcher --values values.yaml .
The default values can be customized by changing them in the helm/values.yaml file.
The created service is of type ClusterIP, so it won't be available publicly. That's why the make portforward command can be run if we wish to access the application locally.
To create a local Kubernetes cluster and deploy the application on it with one command, run make tf_local_apply. This will spin up a new k3d cluster (needs Docker) and installs the Helm chart on it.
Destroy the above created cluster with make tf_local_destroy.