I recently (last week) get the latest rasa-x installed in docker container architecture and using docker compose. Now i am training and modifying the base model using rasa-x from it default directory in docker container (not using git as it is not possible in my case).
My question is - If i want to create and work on 2 different projects, how to do that? Can that be done using same rasa-x installation? or do i need to install rasa/rasa-x again in different folder in docker container? or can use same rasa-x for the creation of new project.
Thanks and Regards
As far as i have remember using Rasa X, the community edition is designed to deploy one bot per instance and thus you would have to replicate the containers in your docker-compose.yml, there is an nginx controller which you can use to point to the Rasa X instance of project X and thus would connect to the Rasa server X.
I think you can train multiple models using the rasa worker instance but to run you would need one container per project instance. because the models are loaded on startup when you run rasa run
I run a complete Rasa-x stack for each Rasa bot.The most annoying thing with this setup is that the nginx server also exists several times (on different ports, which is not convenient). Merging everything into one nginx seemed quite difficult, so I left it as it was.
Lacking multi-project / bot support is a clear downside of the current stack but I think this is where Rasa inc. will step in with a commercial solution at some point
You mean to say… multiple installations of rasa-x in different directories?
Yes, it is one server with several directories (one per stack / chatbot).