Skip to content

Latest commit

 

History

History
94 lines (77 loc) · 4.26 KB

README.md

File metadata and controls

94 lines (77 loc) · 4.26 KB
Raw

Async First Step for FastAPI

This repo is nothing more than instructions for the first level of a FastAPI app. Fork the repo.

Goal

The goal of this repo is to prepare for, and experiment with, a FastAPI fully asynchronous app. Some of the basics have been filled in to both keep things consistent and to alleviate the initial gear-up. Once the basics are established, new development is simple.

Endpoints

To at least somewhat represent asynchronous database lookups and inserts we will need a couple endpoints configured.

GET: /notifications/v1/{notification_id}

This should return the given Notification object or a 404 if it is not found.

POST: /notifications/v1/

This endpoint should take in a payload such as:

    {
        "to": "[email protected]",
        "personalization": {
            "band": "Incubus"
        }
    }

The return should be the Notification object created.

POST: /notifications/v1/multiple

This endpoint should take in a payload such as:

    [
        {
            "to": "[email protected]",
            "personalization": {
                "band": "Korn"
            }
        },
        {
            "to": "[email protected]",
            "personalization": {
                "band": "Soundgarden"
            }
        }
    ]

The return should be the Notification objects created.

Installation Instructions

Since fastapi0 dealt with the Poetry setup this one comes fully-functional. You only need to run:

    poetry install --with static_tools

This will allow your editor to use .venv/bin/python for the interpretter for proper intellisense. You can also enter the poetry shell to perform style checking.

Style Checking

Style checking is defined in pyrpoject.toml and can be executed with the following commands:

    flake8 --select=DCO020 fastapi1
    mypy fastapi1 --strict
    ruff check fastapi1

Alternatively, you can check all at once with ./scripts/run_checks.sh. It is advised that an effort is made to get the checks to pass when the code is complete (maybe starting with the incomplete functions that mypy identifies).

Starting Point

Most of the application is already setup. The fastapi1/status folder contains some example code. The fastapi1/notifications/v1/ folder has an outline to start from.

Fill Outline

The application will start as-is but it is missing any real functionality other than a heartbeat endpoint. Feel free to start it before filling in the outline. To have it fully functioning to store notifications in the DB you'll need to add the missing pieces that would come with building an app of this type. You do not have to follow the outline that is given, but it is recommended.

Container Start

To bring the container up:

    docker compose -f ci/docker-compose-local.yml up

It will start with a single uvicorn worker. You can then follow the URL FastAPI gives you to open the docs. it should be something such as http://localhost:8000/docs#.

Executing the heartbeat GET request should show the following in your docker container:

    app-1  | INFO:     172.18.0.1:57406 - "GET /status/heartbeat HTTP/1.1" 204 No Content
    app-1  | INFO:     GET http://localhost:8000/status/heartbeat 204 0.014161s

Testing

Once the /notifications/v1/ route is established with the required endpoints testing can begin! Much like fastapi0, this will use locust to performance test.

Start the container and in another terminal start locust:

    locust --processes 4

Point the locust host to your app URL. Navigate to the GUI and try 60 users, 10 ramp up, and 30s for the advanced settings (run time).

The locustfile.py has several methods commented out. Try a couple different configurations. Record your results and identify how you can get the most notifications per time interval into the database.

Challenge

Templates and Services were a bit beyond the scope, but would be good to also loop in. If you are up for the challenge bring them into models.py, dao_methods.py, and appropriate routes. Aim for 4NF; the Notifications table is not 4NF currently!