Skip to content

README.md

Latest commit

 

History

History
99 lines (82 loc) · 4.38 KB

README.md

File metadata and controls

99 lines (82 loc) · 4.38 KB

Google Kubernetes Engine (GKE) Deployment

This folder includes the config files for deploying PeerPrep onto GKE environment.

Contents

Requirements

Before starting the deployment, you must install and configure the following tools to create and manage an GKE cluster.

  • kubectl – A command line tool for working with GKE clusters. Guide

  • gcloud – A command line tool for managing Google Cloud resources, including GKE clusters. Guide

You should also have a Google Cloud account and be able to use the Google Cloud Console.

Getting Started

Step 1: Prepare the Docker images

  1. Publish your prepared docker images on the Docker registry.

    You can use the docker-compose.yml in this folder to build the images, just change the target images.

  2. Update the docker images to be used for the main services in each deployment.yaml.

    By default, it will be using our prepared images located in szelongq/peerprep-<SERVICE_NAME>:latest.

Step 2: Create the GKE Cluster

  1. From the Google Cloud Console, choose or create a new project
  2. Enable Kubernetes Engine API for your project
  3. Create an Autopilot cluster using the Google Cloud Console (Refer to: creating-an-auto-pilot-cluster). The Autopilot cluster helps us to manage vertical scaling of resources for us automatically.
  4. Open up your Terminal and login to the glcloud CLI 
    gcloud auth login
  5. Set the project you will be working on 
    gcloud config set project PROJECT_ID
  6. Get authentication credentials and configure kubectl to interact with your cluster 
    gcloud container clusters get-credentials CLUSTER_NAME

Step 3: Create the namespace

  1. Navigate to the k8s folder in the cloned repository
  2. Create the peerprep namespace 
    kubectl apply -f 01-Namespace/.

Step 4: Create the main services

  1. Deploy the services and their endpoints 
    kubectl apply -f 02-Services/.
  2. Check that the deployments and services are visible from the console, or run the following commands: 
    kubectl get deploy
kubectl get services
    There should be 7 deployments and services in total. They might take some time to start up and become available.

Step 5: Add the Horizontal Pod Autoscaler (HPA)

  1. Create the HPAs for the main services

    kubectl apply -f 03-HPA/.

    Take Note: The Matching and Chat Services do not have HPAs due to their reliance on persistent Websocket connections.

Step 6: Updating the ConfigMap

  1. From the Google Cloud Console, navigate to the Services & Ingress tab and check that every service have received a static IP address (You might have to wait for a while).

  2. Update the ConfigMap in 02-Services/00-env-configs.yaml with the new endpoint IP addresses of:

    1. api-gateway-service
    2. matching-service
    3. chat-service

    Take note:

    • REACH_BASE_URL refers to the api-gateway-service endpoint
    • Question, Auth and User service endpoints must also be updated with the base address of api-gateway-service.

  3. Apply the updated ConfigMap

    kubectl apply -f 02-Services/00-env-configs.yaml

Step 7: Access PeerPrep

  1. You should now be able to access Peerprep on the static IP address of the frontend-service.

Helpful Resources