MSI Class Project - Web Application Deployment Automation with Kubernetes

Overview

This project focuses on automating the deployment of a web application for music sharing using Kubernetes. The goal is to create a full-stack application that allows users to register, sign in, upload and download music, as well as delete their uploaded files. The primary aim is to establish a platform for efficient and rapid exchange of audio files. Note that the website is not mobile-friendly yet.

Team Members

• Jaka Pelko
• Miha Vintar

Project Components and key features

1. Kubernetes Deployment Configuration

This repository includes a set of Kubernetes manifests located in the k8s directory. These manifests define the deployment, services, and other resources required for the deployment.

• Nginx Deployment and Service Manifest (nginx.yml)
• Express Deployment and Service Manifest (express.yml)
• Minio StatefulSet, Service and StorageClass Manifest (minio.yml)
• Postgres StatefulSet, Service abd StorageClass Manifest (postgres.yml)
• Ingress Controller Configuration Manifest (ingress.yml)
• ConfigMap Manifest (configmap.yml)
• Secrets Manifest (secret.yml)

2. Persistent Volumes and Persistent Volume Claims

Persistent Volumes and Claims are used for Minio file storage and database storage, ensuring data persistence between pod restarts.

3. Ingress Controller

The Ingress Controller acts as a gateway for managing external traffic.

4. High Availability Deployment

Our goal is to make this application HA (highly available). Nginx and Express are configured to have three instances. This design enhances the reliability of our application, minimizing downtime and offering a robust architecture for uninterrupted user access.

5. Continuous Integration/Continuous Deployment (CI/CD)

Commit of changes in express directory triggers a CI/CD pipeline, established to automate the Docker image build and push on registry.

6. Readiness and Liveness Probes

Our Kubernetes deployment includes configured readiness and liveness probes, designed to enhance the reliability and stability of the application. Configurations are specific for each service based on the complexity, usual startup times and tests we have done (e.g. Nginx has shorter initial delay than Minio since Minio usually requires more time to become operational).

Getting Started

Before proceeding with the steps below, ensure that you have a running Kubernetes cluster, and kubectl is configured to connect to the cluster.

1. Clone the repository:

   git clone git@github.com:MhcVintar/msi-k8s.git

2. Move to the root directory of the repository:

   cd msi-k8s

3. You can now see startScript.sh and deleteScript.sh in current directory. First one handles the deployment of all required objects (services, statefulsets, deployments, etc.) and the second one completely deletes everything related to our project from your computer (deletes PVCs, storageclasses, deployments, services, etc.). Run the following command to automate the app build process:

   ./startScript.sh

4. After script successfully executes, you can check current cluster info with:

   kubectl get all

You should see that multiple containers are setting up (Pending -> ContainerCreating -> Running). After a while all of them are in Ready state.

5. You can now access the website on http://localhost

6. If you want to delete everything K8S related with our project use:

   ./deleteScript.sh

Visual representation (deployment and deletion)

(if you want to access the video instead of gif: https://asciinema.org/a/1ilFV7A0N8mPsc03T6ua8hCZx)

After running startScript.sh and accessing http://localhost, you should see Musify website:

Zero-Downtime upgrades

We have also included a basic example of Rolling update and Blue-green deployment for Nginx service (version 1.0 to version 2.0).

Rolling Update

RollingUpdate.webm

Blue-green Deployment

BlueGreenDeployment.webm