Skip to content

Deployment · By Rivtor · Updated

Deploy Flask with SQLAlchemy and PostgreSQL

Run Flask behind a production WSGI server, read PostgreSQL configuration from the server environment and apply reviewed schema migrations. Disable the development debugger and verify the app’s database operations and static-file handling after release.

See the Rivtor deployment workflow →

Replace the development server

Flask’s local development command is useful while coding but is not the production serving boundary. Choose a supported WSGI server and install it as a runtime dependency. Check whether the application exports app directly or creates it through a factory.

These alternative commands illustrate two project shapes on a Unix-like runtime. Choose the one matching your module and use the port assigned by the host:

gunicorn --bind 0.0.0.0:8000 'app:app'
# Factory-style alternative, not a second process to run:
gunicorn --bind 0.0.0.0:8000 'app:create_app()'

The name app is illustrative. Running from a different service root changes Python’s import resolution. Never enable debug mode to diagnose a public deployment.

Configure the database library deliberately

SQLAlchemy needs a URL compatible with the installed PostgreSQL driver. Keep the full URL private and inspect only sanitized configuration. Set a bounded connection pool; multiply its size by the number of WSGI worker processes before choosing a worker count.

Flask-SQLAlchemy, plain SQLAlchemy and different migration tools have different configuration hooks. Use your actual application’s documented command. Do not run create_all on every request or assume it safely evolves an existing schema. Rehearse migrations on disposable data.

Check the deployed application

CheckExpected observation
StartupProduction WSGI server imports the intended app
HealthThe configured path returns the intended response
Data operationA synthetic record can be written and read
RestartThe record remains in PostgreSQL
Browser pageStatic assets and redirected URLs resolve correctly

Rivtor documents Flask detection and PostgreSQL provisioning. Review the service-root reference and data variables. Behind a proxy, configure forwarded-header handling only for a known trust boundary; blindly trusting arbitrary client-supplied headers can create incorrect URLs or security assumptions.

References