🚀 ML Model Deployment on Docker and Kubernetes

📁 1. Navigate to kubernetes_notes

Move to the 'kubernetes_notes' directory:

cd Deploying-ML-with-FastAPI-on-Docker-Kubernetes

##🐋 2. Deploy on Docker

🔨 2.1. Build Image

Set Docker environment to Minikube

eval $(minikube docker-env)

Build the Docker image

docker image build -t ml-prediction-with-fastapi:1.0 .

▶️ 2.2. Run Container

Run the container

docker run --rm \
  --name ml-prediction \
  -p 8002:8000 -d ml-prediction-with-fastapi:1.0

🔍 2.3. Prediction

Make a prediction

curl -X 'POST' \
  'http://127.0.0.1:8002/prediction/iris' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "SepalLengthCm": 5.1,
  "SepalWidthCm": 3.5,
  "PetalLengthCm": 1.4,
  "PetalWidthCm": 0.2
}'

Expected output: "Iris-setosa"

🛑 2.4. Stop Docker Container

# Stop the running Docker container
docker container stop ml-prediction

More information: https://fastapi.tiangolo.com/deployment/docker/

☸️ 3. Minikube

Apply Kubernetes deployment configuration

kubectl apply -f ml-prediction-deployment.yaml

🔧 3.1. Create Service

Create a NodePort service for the deployment

kubectl create service nodeport ml-prediction --tcp=8000:8000
kubectl get services

Expected output:

NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE kubernetes ClusterIP 10.96.0.1 443/TCP 6h24m ml-prediction NodePort 10.99.159.87 8000:31647/TCP 72s

🌐 4. Prediction on Minikube NodePort

### Make a prediction using the Minikube NodePort service
curl -X 'POST' \
  'http://192.168.49.2:30833/prediction/iris' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "SepalLengthCm": 5.1,
  "SepalWidthCm": 3.5,
  "PetalLengthCm": 1.4,
  "PetalWidthCm": 0.2
}'

Expected output: "Iris-setosa"

🌐 5. Ingress

Enable Ingress addon in Minikube

minikube addons enable ingress

Apply the Ingress resource

kubectl apply -f ingress-ml-prediction.yaml

Add the following line to /etc/hosts

sudo vim /etc/hosts

#### 192.168.49.2    ml.prediction.vbo.local

Test the prediction through Ingress

curl -X 'POST' \
  'http://ml.prediction.vbo.local/prediction/iris' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "SepalLengthCm": 5.1,
  "SepalWidthCm": 3.5,
  "PetalLengthCm": 1.4,
  "PetalWidthCm": 0.2
}'