SQLAlchemy / Flask / PostgreSQL pool connection

Question

After having played for a long time with Django, I\'m trying a bit of Flask with SQLAlchemy, and I must say I quite like it. However there is something that I don\'t get: I have

Answer 1

Flask-SQLAlchemy creates a SQLAlchemy engine using the create_engine method in SQLAlchemy, which you can read about some of the options and defaults in the documentation for the create_engine function. According to the Flask-SQLAlchemy documentation, you can specify some of the configuration options specific to pooling. You can set those values in various ways, which you can read about in the Flask configuration. You already have a settings configuration module, so you can add to your config file something like...

SQLALCHEMY_POOL_SIZE=10

So, yes, there is automatic connection pooling. This is provided by SQLAlchemy by default. As of the posting of this answer, Flask-SQLAlchemy allows you to modify some of these options using the configuration file (although it appears there is an old pull request to allow you to specify ANY create_engine parameter).

If you need more support for configuring the SQLAlchemy engine than Flask-SQLAlchemy provides, you can either use SQLAlchemy without the Flask-SQLAlchemy wrapper, or modify Flask-SQLAlchemy (perhaps merging the pull request) to allow this.

Answer 2

Every worker of your flask instance running a proper wsgi server with multiple workers will have it's own engine - because your script is loaded for every worker separately.

In consequence you have simply no control about the real pooling, except you limit your worker threads of wsgi server and calculate the max allowed connections per engine/pool instance.

Further details can be read at https://docs.sqlalchemy.org/en/13/core/pooling.html#using-connection-pools-with-multiprocessing

Answer 3

NullPool can be set on the create_engine method

` from sqlalchemy.pool import NullPool
  engine = create_engine('db connection url',poolclass=NullPool)`