2-function-apps.md

Deploy the Function Apps

Once the infrastructure has been deployed, you will need to deploy three Function Apps that implement the platform's data pipelines. Two of these are written in Python:

• func_trigger
• func_dataproc

And one is written in .NET/C#:

• func_extract

Although written in different languages, the steps to deploy them are similar, using the Azure command line tools.

Pre-requisites

You will need the following installed on your local machine:

• Docker, i.e. the docker command. This will be used to build the Python apps locally. In most cases, you will use Docker Desktop.
• Azure CLI, i.e. the az command. See How to install the Azure CLI.
• Azure Functions Core Tools, i.e. the func command. See Install the Azure Functions Core Tools
• The .NET SDK, i.e. the dotnet command. See Install .NET on Windows, Linux, and macOS.

You will run the commands below from a Unix-like shell terminal session, using a variant of sh or bash.

Before running the deployment commands, make sure you are logged in to your Azure subscription using az login. You can check which subscription you are connected to with az account show.

Deploy the Python apps

Change to the target app directory:

cd src/python/functions/func_trigger

We must first install all the Python requirements for the app. Since the app is executed on a Linux host, we need to execute this step in a Docker container running Linux, so that we get the Linux versions of any binary dependencies. We will use the standard Python image for this. Launch the container locally using this command:

docker run --platform linux/amd64 --rm -it -v $(pwd)/../..:/app amd64/python:3.8 bash

This command will place you in a shell running in Docker container, where you can run the following steps. The -v option is used to mount the source code directory from your machine to the container.

Once in the container, change to the source code directory:

cd /app/functions/func_trigger

Install the required Python dependencies for the function app. This will not install anything in the system Python environment, it will install the dependencies in a sub-directory specific to the function app, using the --target option of pip install.

pip install --target=".python_packages/lib/site-packages" -r ../../common/requirements.txt -r requirements.txt
pip install --target=".python_packages/lib/site-packages" ../../common

You may get some warnings when installing these dependencies; they can be ignored.

You can now exit the container to run the rest of the steps.

Deploy the function app using the Azure Functions Core Tool.

func azure functionapp publish <function app name> --python --no-build

You can find the function app name in the outputs of the Terraform run; it should look like this: functrigger-stg-xxxx.

Repeat the same steps for the func_dataproc function app:

• Change to the src/python/functions/func_dataproc folder.
• In the container, go to the /app/functions/func_dataproc directory.
• Use the same pip install commands.
• When deploying, the function app name will look like: funcdataproc-stg-xxxx.

Deploy the .NET app

Change to the target app directory:

cd src/dotnet/functions/func_extract

Build the app using the .NET CLI:

dotnet restore
dotnet build --configuration Release

Deploy the function app using the Azure Functions Core Tool.

func azure functionapp publish <function app name> --dotnet