Deployable Fullstack application into GKE

Simple application that consists of three services for highlighting running containerized applications on Google Cloud.

• app - Web frontend (Angular app)
• api - API microservice (Nest.js app)
• database - Backend (Postgres SQL database)

Quickstart (GKE with Autopilot)

Prerequisites

• Install Google Cloud SDK (gcloud)
• Install Skaffold

Create an Autopilot cluster

With an Autopilot cluster, GKE manages compute infrastructure for Kubernetes, freeing developers from physical cluster administration to focus on building and maintaining applications.

You will need a Google Cloud project with billing enabled.

Follow either of the following instructions to create an Autopilot cluster:

# Export variables 
export PROJECT_ID=fullstask-k8s
export REGION=europe-southwest1
export CLUSTER_ID=fullstack-k8s-cluster

# Authorize gcloud
gcloud auth login

# Create Autopilot cluster
gcloud container clusters create-auto $CLUSTER_ID \
--project $PROJECT_ID \
--region $REGION \
--monitoring=SYSTEM,WORKLOAD

Build and deploy the app

You will need to install Skaffold on your system. This tool uses a file (./skaffold.yaml) that configures the build and deploy tasks for the application:

• Application components will be built on Google Cloud using Cloud Build so that you won't need to install any other tools, including Docker, on your own system.
• The application will be deployed to the GKE Autopilot cluster you created previously.

1. Install Skaffold on your system

Install Skaffold for CLI.

2. Use scaffold to run the deployment

Ensure the following environment variables are set:

• PROJECT_ID set to your project
• REGION set to the region
• CLUSTER_ID set to your cluster

export PROJECT_ID=fullstask-k8s
export REGION=europe-southwest1
export CLUSTER_ID=fullstack-k8s-cluster

Get authorization to access your project.

gcloud auth login

Get Kubernetes credentials for the GKE cluster in your project.

gcloud container clusters get-credentials \
--project $PROJECT_ID \
--region $REGION \
$CLUSTER_ID

Run the following skaffold command. You may be prompted to enable various APIs, such as Cloud Build.

skaffold run --default-repo=gcr.io/$PROJECT_ID/fullstack-k8s --tail

Test the app

Print the public address of the app.

echo "http://"$(kubectl get service web-external -o jsonpath="{.status.loadBalancer.ingress[0].ip}{'\n'}")

Skaffold displays streaming log output in the terminal.

Note: you can also get the public address from the External endpoints field from this link (substituting values for REGION, CLUSTER_ID and PROJECT_ID) https://console.cloud.google.com/kubernetes/service/REGION/CLUSTER_ID/default/web-external/overview?project=PROJECT_ID

Clean up

Press Ctrl-C to stop Skaffold log streaming.

To stop the voting app from running and to delete Kubernetes resources in the cluster, enter:

skaffold delete

Local development & testing

Prerequisites

1. Install Docker.
2. Clone this repo and change directory to it.

git clone https://github.com/subfuzion/voting-demo.git
cd voting-demo

You will execute all commands in this from the top-level directory of the repo.

Run the complete application

docker compose up

Wait until you see the following log entry in the terminal output:

fullstack-k8s-app-1       | INFO:waitress:Serving on http://0.0.0.0:8080

Open a brower and navigate to http://localhost:8080

When finished, run:

docker compose down --remove-orphans

Test the database package

The database package is a Node.js package for working with the database. It's meant to be used by the api microservice. It's published to npm as @subfuzion/vote-database.

To test it:

docker compose -f tests/docker-compose.postgres.test.yaml run sut

After making changes to the package source code, run:

docker compose -f tests/docker-compose.postgres.test.yaml down --remove-orphans
docker compose -f tests/docker-compose.postgres.test.yaml build
docker compose -f tests/docker-compose.postgres.test.yaml run sut

When finished, run:

docker compose -f tests/docker-compose.postgres.test.yaml down

Test the API service

docker compose -f tests/docker-compose.vote.test.yaml run sut

After making changes to the package source code, run:

docker compose -f tests/docker-compose.vote.test.yaml down --remove-orphans
docker compose -f tests/docker-compose.vote.test.yaml build
docker compose -f tests/docker-compose.vote.test.yaml run sut

When finished, run:

docker compose -f tests/docker-compose.postgres.test.yaml down

Partially containerized dev environment

Sometimes it's convenient to run some services, such as the database, in a container, while editing/testing/debugging other services uncontainerized.

Note: for any of the docker compose up commands, you can run services detached by appending the -d option. To view log output, run docker compose logs (with the -f option to stream output). To run the database in a container:

docker compose up postgres

To run both the database and api services in containers:

docker compose up postgres api

When running an uncontainerized version of either the vote or the web services, you will need to set certain environment variables so the services know how to connect to their upstream dependencies:

To run the api service uncontainerized:

export PGHOST=localhost
(cd src/api; npm run)

To run uncontainerized api tests:

export API_HOST=localhost;
API_PORT=8080
(cd src/api; npm test)

To run the web service uncontainerized:

export HOST=localhost
export PORT=5000
export API=http://localhost:8080
(cd src/app, ng serve)