Rasa SQL DB configuration

Hi Team,

I am looking forward to integrate the rasa chatbot with psycopg2 as dialect for SQL. I am not getting enough content to start with. Can you please provide me some sample examples of endpoint.yml file wherein SQL is used as Tracker Store.

Thanks for your continuous support.

You should just need something like this:

  store_type: SQL
  url: localhost
  db: rasa
  dialect: "postgres"

Let me know if something like this doesn’t work

Hi For the record, working configuration for me on RASA 1.9.4 with PostgreSQL in docker also (everything launched through custom docker-compose.yml, all connected on the same network so dockers can called themselves by their service names) is the following :

  store_type: SQL
  dialect: "postgresql"
  url: "postgresql://<login>:<pass>@<service_name_of_postgres_docker>:5432/<db_name>"

Indifferently postgres or postgresql as a keyword for dialect and url. Seemed that other environment variables didn’t make it work (db name, login and password all apart), I had to put everything into the url (a bit like the link provided towards SQLAlchemy website -PostgreSQL — SQLAlchemy 1.3 Documentation except that psycopg2 additional parameter to postgres doesn’t seem relevant).

One thing annoying from a “ops perspective” : RASA seems to need a working trained model to be able to dump even the very first lines into database. If there is no model provided, the user entries are not dumped, even if RASA sees it. So just gluing all technical stacks together is not enough, we need to provide a minimum conversation to test everything.