briancappello / eiq-demo-app

The App

The app is a small Flask app structured using the application factory pattern. It uses Flask-WTF and Pandas to handle the form and validation requirements. Due to time constraints, the templates use Jinja2 and Bootstrap4 for styling (I reused some open source code I had already written for the layout and utility macros).

Live on GCP (I didn't set up any DNS records or HTTPS)

GKE and Docker commands

This is where the bulk of my time went. I had toyed around with GCP and Kubernetes briefly maybe 2 years ago, but that being my only experience with them, I was essentially getting set up from scratch. I found the GCP console and related documentation to be a bit of a labyrinth, but after locating the relevant docs it wasn't bad to get "hello world" deployed. I faced a little bit of a hiccup debugging gunicorn with the transition from a single file to using the application factory pattern, which I found much easier to solve building and running docker images locally than by using Google's push build service.

# install google cloud SDK, kubectl, and docker

# set env vars
export PROJECT_ID="eiq-demo"
export GCP_REGION="us-west1"
export GCP_ZONE="us-west1-a"
export REPO_NAME="$PROJECT_ID-repo"  # for GCP Artifact Registry
export IMAGE_NAME="$PROJECT_ID-app"  # for docker image name stored in Artifact Registry
export CLUSTER_NAME="$PROJECT_ID-cluster"

# configure local gcloud env and authenticate
gcloud config configurations create $PROJECT_ID
gcloud config set project $PROJECT_ID
gcloud auth login
gcloud auth application-default login  # instead of setting up a service acct

# create an artifact repo to store containers
gcloud artifacts repositories create $REPO_NAME \
    --project=$PROJECT_ID \
    --repository-format=docker \
    --location=$GCP_REGION \
    --description="Docker repository"

# build & push the docker file
gcloud builds submit \
  --tag $GCP_REGION-docker.pkg.dev/$PROJECT_ID/$REPO_NAME/$IMAGE_NAME .

# create a GKE cluster
gcloud container clusters create $CLUSTER_NAME --num-nodes 1 --zone $GCP_ZONE

# verify access to the cluster
kubectl get nodes

# create deployment.yaml and deploy it
kubectl apply -f deployment.yaml

# create service.yaml and deploy it
kubectl apply -f service.yaml

# list services to get EXTERNAL-IP
kubectl get services

# pushing new code
# ----------------

# rebuild container with a new tag
gcloud builds submit \
  --tag $GCP_REGION-docker.pkg.dev/$PROJECT_ID/$REPO_NAME/$IMAGE_NAME:v2 .

# set the container the GKE deployment should use
kubectl set image deployment/app-deployment \
  app-container=$GCP_REGION-docker.pkg.dev/$PROJECT_ID/$REPO_NAME/$IMAGE_NAME:v2

# building locally and pushing
# configure docker auth: https://cloud.google.com/artifact-registry/docs/docker/authentication#standalone-helper

docker build -t localtag:v1 .
docker tag localtag:v1 $GCP_REGION-docker.pkg.dev/$PROJECT_ID/$REPO_NAME/$IMAGE_NAME:local
kubectl set image deployment/app-deployment \
  app-container=$GCP_REGION-docker.pkg.dev/$PROJECT_ID/$REPO_NAME/$IMAGE_NAME:local

Testing using pytest

Tests are run locally, using

pip install -r requirements-dev.txt
FLASK_ENV="test" pytest

(ideally these would be configured to run inside docker using tox, but I used up my allotted time)

Approx time investment

Task	Time
Development environment setup	.5hrs (installing and configuring Cloud SDK and Docker)
Hello World to first live deploy on GCP	2.5hrs
Writing the app	1.5hrs
Writing tests	.5hrs
Debugging prod issues	.5hrs
Manual testing and documentation	.5hrs
Total	6hrs