I have noticed that running

cursor.execute

with the SqlAlchemy block doesn't persist inserts. The block can connect to the db just fine and even fetch records just fine. But it can neither do inserts nor updates. FWIW, I'm executing from a jinja template like so:

cursor.execute(insert_records.render())

Which database driver/dialect are you using?

Psycopg2

Hmm, that's strange. SQLAlchemy normally auto-commits when using the psycopg2 driver. Does it work if you use the block's execute method instead of using the cursor directly?

Yes, I'm using the execute method based on my example above

Ah, got it. You're using the connection directly. Does it work if you do this instead?

@flow
def contracts(network: str):
    logger = get_run_logger()

    connector: SqlAlchemyConnector = SqlAlchemyConnector.load(network)
    <http://logger.info|logger.info>(f"connected to sqlalchemy db: {network}")
    <http://logger.info|logger.info>("updating contracts rollup")
    connector.execute(update_contracts_tpl.render())

That was the first method I used actually and that also did not work.

Does it work if you tell it to commit?

@flow
def contracts(network: str):
    logger = get_run_logger()

    connector: SqlAlchemyConnector = SqlAlchemyConnector.load(network)
    with connector.get_connection() as c:
        <http://logger.info|logger.info>(f"connected to sqlalchemy db: {network}")

        <http://logger.info|logger.info>("updating contracts rollup")
        c.execute(update_contracts_tpl.render())
        c.commit()

It should auto commit - but if this works, it will give useful feedback on what's going wrong.

Alright, let me try that