Skip to content

Latest commit

 

History

History
257 lines (179 loc) · 7.58 KB

DEV.md

File metadata and controls

257 lines (179 loc) · 7.58 KB

Developer Setup

Dependencies

Install Docker

Install Docker Desktop: https://docs.docker.com/get-docker/

Frontend

How to setup

  1. Set up local environment variables

    For local dev, create a .env file at the root of the frontend/ directory using the .env.example file:

    cp .env.example .env

  2. Install the dependencies.

    npm install

  3. Run.

    npm run develop

Backend

Set up local environment variables

For local dev, create a .env.local file at the root of the backend/ directory using the .env.example file:

cp .env.example .env.local

Setup the application and database

NOTE: You must pass an ENVIRONMENT variable to ensure you're running in the right env.

  1. Run the app using docker compose:
ENVIRONMENT=local docker compose up backend database --build

OR

./run.sh
  1. Run all the alembic migrations to load the schema:
ENVIRONMENT=local docker compose exec backend alembic upgrade head
  1. Seed the local database:
ENVIRONMENT=local docker compose exec backend python -m scripts.seed_db
  • This will add a fake user with a language code spa and country code ES, then create some tasks the user can edit / submit.

Adding alembic migrations

Alembic Docs

To add a migration run:

alembic revision -m "some_revision_name"

Make sure you add an upgrade() and downgrade() step. Once your migration is added, rebuild the backend container, and test that we can upgrade and downgrade a couple times without the DB getting into a broken state:

ENVIRONMENT=local docker compose exec backend alembic upgrade head
ENVIRONMENT=local docker compose exec backend alembic downgrade -1

To check the history of migrations run:

ENVIRONMENT=local docker compose exec backend alembic history

Interacting with the database container

The docker compose command will spin up the backend along with a postgres container that you can interact with.

To connect to the DB inside of docker, you can run:

ENVIRONMENT=local docker compose exec database psql instruct_multilingual -U backendapp

and interactively query data in there such as:

select * from language_code;

Modifying leaderboard jobs

We currently have some Cloud Run Jobs that run on a scheduled basis via Cloud Scheduler. Each of these jobs runs in their own container, and live in the directory backend/jobs/, with each type of job in their own subdirectory like daily/ and their own Dockerfile with a command they run.

They use the leaderboard_update_job.py script and have the same dependencies as the FastAPI server.

Deploying jobs

To deploy and modify the schedule of the jobs, you can use the ops script:

ops/create-jobs.sh

Which will tell you the options you can use. You must pass a job type and a cron expression as a schedule.

Running jobs locally

You can run the leaderboard jobs locally with:

ENVIRONMENT=local docker compose up <job_type>_leaderboard_job --build

replacing <job_type> with the available options (see the docker-compose.yaml file).

Run unit tests

Currently, we have a .env.test file that gets read in by unit tests. To run the unit tests, you'll need a local postgres database or run unit tests in the GitHub Actions workflow.

Recommended: Run the dockerized postgresql instance using:

docker compose up database

then run unit tests, which will use the dockerized server and a database called testdb

ENVIRONMENT=test pytest

Run in staging mode

For connecting to the staging database, credentials should be retrieved from GCP Secret Manager, or via a secure message channel.

  1. Create a file called .env.staging in the root of the backend/ project:
cp .env.example .env.staging
  1. Retrieve the staging database URL from GCP Secret Manager
  2. Retrieve the staging discord keys from GCP Secret Manager
  3. Replace the values in the .env.staging file with the staging values
  4. Remove POSTGRES_DB, POSTGRES_USER, POSTGRES_PASSWORD fields
  5. Start the server with ENVIRONMENT=staging:
ENVIRONMENT=staging docker compose up backend --build

Run in production mode

NOTE: Only run in production mode for read-operations and debugging.

For connecting to the production database, credentials should be retrieved from GCP Secret Manager, or via a secure message channel.

  1. Create a file called .env.production in the root of the backend/ project:
cp .env.example .env.production
  1. Retrieve the production database URL from GCP Secret Manager
  2. Retrieve the production discord keys from GCP Secret Manager
  3. Replace the values in the .env.production file with the production values
  4. Remove POSTGRES_DB, POSTGRES_USER, POSTGRES_PASSWORD fields
  5. Start the server with ENVIRONMENT=production:
ENVIRONMENT=production docker compose up backend --build
  • This will connect to the production database so you don't need to run the local database container

Deploying services to GCP Cloud Run (production)

To deploy the frontend / backend to GCP Cloud Run (production) there are a few involved steps. Namely:

  • Building images
  • Pushing to Container Registry
  • Deploying the latest image to Cloud Run

This is automated via GCP Cloud Build. cloudbuild-frontend.yaml will automatically do the above steps if there are changes to the frontend/ directory, and cloudbuild-backend.yaml will do the same if there are changes to the backend/ directory.

Aya Analytics App

Set up local environment variables

For local dev, create a .env.analytics-instruct-multilingual-app.local file at the root of the analytics/ directory and add the following environment variables to it:

INSTANCE_CONNECTION_NAME="" # Cloud SQL instance connection name
DB_USER="" #user name to access DB
DB_PASS="" #password for DB
DB_NAME="" #name of the DB
C4AI_PROJECT_ID="" #C4AI GCP project ID
GRADIO_SERVER_NAME="0.0.0.0"
GRADIO_SERVER_PORT=8080
APP_ENVIRONMENT="local"

Install Requirements

Create a python virtual environment and install the necessary libraries using requirements.txt file

pip install -r requirements.txt

Run the application

Run the app using app.py:

python app.py

OR

You can also run the app in reload mode using below command:

gradio app.py