Temperature reader

Environment description

• Houses have rooms and rooms can have lights that can be controlled individually
• Each house has a furnace. The furnace can either be
  • off,
  • fan-on,
  • heat
• Each room can have sensors used for monitoring.
• Sensors can be for indor temperature, humidity, outdoor temperature. For this task only indor temperature task matter.
• Sensors are linked to a 3rd party API that exposes the latest sensor reading every 5 minutes, the historic data for each of these sensors is NOT stored on the remote source
• Sensor API https://api-staging.paritygo.com/sensors/api/sensors/

Requirements for API

• get information about a given house and rooms;
• change the furnace state;
• turn lights on and off;
• view historic sensor data for the past 15 minutes.

Assumptions

• App is running in the house itself to have access to the furnace and lights switchers
• It means that hardware resources are limited
• Intensive use of network connections, to store / retrieve data to / from a cloud based database may or may not be the option
• Any services and scripts which are supposed to sync / backup / clean sensor data history, loggs or other stored info are out of scope for the test purpose.
• A sensor type has only one metric to measure (according to the API provided).
• Sensors and lights can be outside the house, we put them into 'outside' room.
• We maintain sensor slugs to be equal to slugs from the third party API. If a new sensor comes in API, we'll see it in the errors log. Mapping of sensor slugs is out of scope.

Architecture

• 2 services to separate concerns:
  • Home API - Django ORM, Django*REST
    • get information about a given house and rooms;
    • change the furnace state;
    • turn lights on and off;
    • view historic sensor data for the past 15 minutes.
  • Sensors Poll - Requests, Celery, Redis
    • fires every 5 min
    • polls sensor API
    • calls Home API to store sensor data
• SQLite is used for a local database, because of described assumptions
• Daily, another service should sync any data, we need from our local snapshot db, to the cloud based db and clear local db. According to made assumptions, it's out of scope.
• Historic sensor data always shows the last data obtained from sensors API and all data which is 15 min older than the last loaded data. It matters, if data is not inserted every 5 min.
• Auth:
  • 3 types of auth are enabled: JWT, Session, and Basic; Basic enabled only for testing API with a browser;
  • JWT refresh is not used to simplify client and do not persist / refresh token (which could've been another celery task)

Setup and run

Both apps home and sensors_poll have requirements.txt files to run installation.

All files outside apps folders are required only for development. For example, requirements.dev.local.txt has contains dependencies like pylint and yapf.

home app

1. Run migration pythom manage.py migrate
2. Load fixture python manage.py loaddata ./api/fixtures/house-plan.yaml
3. Create admin user (any name is okay) python manage.py createsuperuser --email admin@example.com --username admin
4. Run server python manage.py runserver
5. Visit http://localhost:8000/admin/ for admin interface, use the user you created to log in
6. Visit http://localhost:8000/api/docs/ to explore Home API
7. You can check also http://localhost:8000/api/schema/ and start testing at http://localhost:8000/api/
8. When done, stop server with Ctrl+C

sensors_poll

1. You need to have redis running on port 6379 to be able to run celery.
2. Rename poll.ini.example to poll.ini, open it and insert there username and password you chose when set up home app
3. Start beat and worker from the directory which contains sensors_poll folder with command celery -A sensors_poll worker -B --loglevel=info (feel free to change loglevel)
4. When done, stop celery with Ctrl+C