Machine Learning Predict at Cloud. This project uses MS Azure PaaS and it´s a DevOps compliance project.

This project is designed to be deployed at MS Azure App Services, using DevOps CI/CD pipelines

• A link to a Trello board for the project ML WebApp Project Board
• A link to Project Plan spreadsheet: ML WebApp Project Plan

1. Clone this repository
2. Create your infrastructure as code
3. Update this azure_pipelines.yml to reflect your Azure environment.

1. Create an Azure Account
2. Install the Azure command line interface
3. Create an Azure DevOps account
4. Install Azure Cloud Shell

• Architectural Diagram
  • 
  • 

PRE-REQ: Make sure you are logged at your Azure via Azure CLI command:

az login

1. Clone this Github repository locally:

git clone https://github.com/Pet-slack/udacity-cicd-demo.git
cd udacity-cicd-demo

2. Create your Python virtual environment

• Create a Python3.x virtualenv

python3 -m venv ~/.v-python

• Activate your virtualenv

source ~/.v-python/bin/activate

3. Run your project locally

• Set it up:

make all

• Run the program:

python app.py

• Test the program running a ML prediction:

sh ./make_prediction.sh

• Check Github Actions CI status

1. Initial project App Service setup/deploy:

az webapp up --name <APP_NAME> --sku F1 --location <LOCATION> --resource-group <RESOURCE_GROUP>

2. Setup and deploy CI/CD 💪

• Log in at Azure DevOps and setup a new project using this Github Repo as source
• Create a Project: New Project >> Go to Pipelines: Create Pipeline >> Select Project Repository >> Save
• Set up a new automatic Service Connection called myUdacitySP_Conn in Project Settings >> Service Connections >> Azure Resource Manager
• Deploy the DevOps project to Azure App Service running the CI/CD pipeline

✔️   Run Pipeline: Go to Project >> Pipelines >> Pipelines >> Click on Pet-slack.udacity-cicd-demo >> Click on Run pipeline

3. Check the CI/CD deployment 💙

4. Test Azure App Service deployment:

udacity@Azure:~$ ./make_predict_azure_app.sh
Port: 443
{"prediction":[20.35373177134412]}

5. Output of streamed log files from deployed application

az webapp log tail --name <APP_NAME> --resource-group <RESOURCE_GROUP>

Run the following commands to fully deploy this project at MS Azure.

curl -s https://raw.githubusercontent.com/Pet-slack/udacity-cicd-demo/master/scripts/full-deploy-commands.sh | bash -s -- start
curl -s https://raw.githubusercontent.com/Pet-slack/udacity-cicd-demo/master/scripts/full-deploy-commands.sh | bash -s -- deploy

Run the following command to delete the Azure App Service.

curl -s https://raw.githubusercontent.com/Pet-slack/udacity-cicd-demo/master/scripts/full-deploy-commands.sh | bash -s -- delete

Locust load testing output for Azure App Service

Locust Test based on scripts/locustfile.py

• Add multiple stages:
  • Deploy CD DEV stage. In progress at ml-dev branch
  • Deploy CD Github Actions option. Pending

Yutube Demo