vikrantarora14 / module4projectml Goto Github PK

You are given a pre-trained, sklearn model that has been trained to predict housing prices in Boston according to several features, such as average rooms in a home and data about highway access, teacher-to-pupil ratios, and so on. This project is to operationalize a Python flask app—in a provided file, app.py—that serves out predictions (inference) about housing prices through API calls.

1. app.py - the main python file which has logic for making predictions
2. make_predictions.sh - calls app.py to trigger predictions
3. .circleci - is a hidden folder which inturn contains config.yml for running circleci build
4. requirements.txt - necessary components to be installed for the project to run successffully
5. Makefile - you can trigger the basic setup via Makefile such as linting as well as installing all components
6. run_docker-sh - creates docker image locally
7. upload_docker.sh - uploads the docker image to dockerhub
8. run_kubernetes.sh - launches the docker image in a kubernetes cluster
9. output_txt_files contains two files: docker_out.txt - this contains the execution logging information when app.py is triggered via make_predictiions.sh kubernetes_out.txt - this contains the logs of the steps executed when ./run_kubernetes.sh is executed.

• Create a virtualenv and activate it *python3 -m venv ~/Desktop/CloudDevOps/Microservices/Project/.devopsprojectmlmicroservicekubernetes

*source python3 -m venv ~/Desktop/CloudDevOps/Microservices/Project/.devopsprojectmlmicroservicekubernetes/bin/activate

• Run make install to install the necessary dependencies make install

• Run make lint to lint app.py and Dockerrfile make lint

1. Standalone: python app.py
2. Run in Docker: ./run_docker.sh
3. Run in Kubernetes: ./run_kubernetes.sh

*Execute "./run_docker.sh" *this will create docker image locally and the application will be up and running on port 8000

*Execute "./make_prediction.sh" *this will make a call to app.py and the logs that are generated for the steps executed in app.py go into docker_out.txt

./upload_docker.sh

./run_kubernetes.sh *This will use the image that was uplaoded onto Dockerhub and launch it in a pod called projectml in kubernetes cluster *The applicatiion will run on port 8000

*Once the application is up and running on port 8000 once ./run_kubernetes.sh is successful, trigger ./make_prediction.sh to make a call to the application to return predictions.

*The logs are then copied to kubernetes_out.txt