Use gunicorn in docker for deployment

Author: lopezco

README.md

@@ -44,9 +44,7 @@ or one of the scripts in the `./example` folder
 
 ### On Docker
 
-#### Development
-
-Build the image (this has to be done every time the code or the model change)
+Build the image (this has to be done every time the code or the model changes)
 ```bash
 $ docker-compose build
 ```
@@ -55,19 +53,6 @@ Create and run the container
 $ docker-compose up
 ```
 
-#### Production
-
-Using uWSGI and nginx for production.
-
-Build the image (this has to be done every time the code or the model change)
-```bash
-$ docker-compose -f docker-compose-production.yml build
-```
-Create and run the container
-```bash
-$ docker-compose -f docker-compose-production.yml up
-```
-
 ### On local Python environment
 
 Create the environment
@@ -137,6 +122,8 @@ Endpoint: `/service-info`
 
 ```bash
 $ curl -X GET http://localhost:5000/service-info
+```
+```json
 {
   "debug": true,
   "running-since": 1563355369.6482198,
@@ -152,6 +139,8 @@ Endpoint: `/info`
 
 ```bash
 $ curl -X GET http://localhost:5000/info
+```
+```json
 {
   "metadata": {
     "features": [
@@ -191,6 +180,8 @@ Endpoint: `/predict`
 
 ```bash
 $ curl -d '[{"feature1": 1, "feature2": 1, "feature3": 2}, {"feature1": 1, "feature2": 1, "feature3": 2}]' -H "Content-Type: application/json" -X POST http://localhost:5000/predict
+```
+```json
 {
   "prediction": [0, 0]
 }
@@ -202,6 +193,8 @@ Endpoint: `/predict?proba=1`
 
 ```bash
 $ curl -d '{"feature1": 1, "feature2": 1, "feature3": 2}' -H "Content-Type: application/json" -X POST "http://localhost:5000/predict?proba=1"
+```
+```json
 {
   "prediction": [{
     "0": 0.8,
@@ -216,6 +209,8 @@ $ curl -d '{"feature1": 1, "feature2": 1, "feature3": 2}' -H "Content-Type: appl
 Endpoint: `/features`
 ```bash
 $ curl -X GET "http://localhost:5000/features"
+```
+```json
 [
   {
     "default": -1,
@@ -243,7 +238,9 @@ $ curl -X GET "http://localhost:5000/features"
 Endpoint: `/predict?proba=1&explain=1`
 
 ```bash
-$curl -d '{"feature1": 1, "feature2": 1, "feature3": 2}' -H "Content-Type: application/json" -X POST "http://localhost:5000/predict?proba=1&explain=1"
+$ curl -d '{"feature1": 1, "feature2": 1, "feature3": 2}' -H "Content-Type: application/json" -X POST "http://localhost:5000/predict?proba=1&explain=1"
+```
+```json
 {
   "explanation": {
     "feature1": 0.10000000149011613,

docker-compose-production.yml

@@ -10,4 +10,4 @@ services:
       - ./docker/.env
     restart: always
     ports:
-      - "33333:5000"
+      - "5000:5000"

docker-compose.yml

@@ -1,3 +1,2 @@
-DEBUG=1
 MODEL_NAME=model.joblib
 MODEL_TYPE=SKLEARN_MODEL

docker/.env

@@ -1,6 +1,8 @@
 FROM python:3
 
-WORKDIR /usr/src/app
+WORKDIR /app
+
+RUN pip install --upgrade pip
 
 # Copy and install service requirements
 COPY requirements-service.txt .
@@ -13,5 +15,7 @@ RUN pip install -r ./requirements.txt
 # Copy code and model
 COPY . .
 
-ENTRYPOINT ["python"]
-CMD [ "./service.py" ]
+EXPOSE 5000
+
+CMD gunicorn -b 0.0.0.0:5000 service --timeout 300 --workers=2 --threads=4 --worker-class=gthread
+

docker/Dockerfile

@@ -1,5 +1,6 @@
 # Web Service
 flask
+gunicorn
 
 # Model persistance and data handling
 numpy

docker/Dockerfile-production

@@ -2,3 +2,4 @@ pandas
 numpy
 scikit-learn
 scipy
+lightgbm

docker/production.env

@@ -15,7 +15,7 @@
 # Version of this APP template
 __version__ = '2.1.0'
 # Read env variables
-DEBUG = os.environ.get('DEBUG', True)
+DEBUG = os.environ.get('DEBUG', False)
 MODEL_NAME = os.environ.get('MODEL_NAME', 'model.joblib')
 ENVIRONMENT = os.environ.get('ENVIRONMENT', 'local')
 MODEL_TYPE = os.environ.get('MODEL_TYPE', 'SKLEARN_MODEL')