Rasa X 1.0.0 Training error

Hello everyone,

I’m pretty excited about the new Rasa X version. The installation is very easy and smooth, however I seem to have a problem to train anything on it and I’m looking for your insights.

I made a clean install of Ubunutu 20.04 Desktop I am running as a VM.

I installed Rasa Open Source locally with Python and here’s the information the command rasa --version is giving me.

Rasa Version      :         2.8.21
Minimum Compatible Version: 2.8.9
Rasa SDK Version  :         2.8.3
Rasa X Version    :         None
Python Version    :         3.8.10
Operating System  :         Linux-5.11.0-44-generic-x86_64-with-glibc2.29
Python Path       :         /home/cio-rasa/rasa/venv/bin/python3

rasactl --version

rasactl version 0.5.0

rasactl status

Name:                   	eager-blackburn                         	
Status:                 	Running                                 	
URL:                    	http://eager-blackburn.rasactl.localhost	
Version:                	1.0.0                                   	
Enterprise:             	inactive                                	
Rasa production version:	0.0.0                                   	
Rasa worker version:    	0.0.0                                   	
Project path:           	not defined     

I did use rasactl connect rasa in a different terminal and it gives me this

(production-worker) 2022-01-09 11:49:02 INFO     root  - Starting Rasa server on http://localhost:5005
(production-worker) 2022-01-09 11:49:02 INFO     rasa.core.brokers.pika  - Connecting to RabbitMQ ...
(production-worker) 2022-01-09 11:49:02 INFO     rasa.core.brokers.pika  - RabbitMQ connection to '127.0.0.1' was established.
(production-worker) 2022-01-09 11:49:02 INFO     root  - Rasa server is up and running.

When I try to train a model Rasa X tells me it failed and rasactl logs gives me the following error

[2022-01-09 16:50:02 +0000] - (sanic.access)[INFO][10.244.0.8:54496]: GET http://eager-blackburn.rasactl.localhost/api/projects/default/models/tags/production?token=rasaXToken  404 159
ERROR:rasax.community.services.stack_service:Cannot connect to host gateway.docker.internal:5005 ssl:default [Name or service not known]
Encountered an exception while training. Please check the logs of the rasa worker container for more information.

I’m a bit confused as I understood the docker install was deprecated. Anyone can shed some lights here?

Hey @mrsaboteur , the new REI installation method makes use of Kind - it runs a local Kubernetes cluster by using Docker containers as “nodes”. As stated in the documentation, docker is a requirement to be able to run rasactl. Note that this is not the same as using docker container or docker-compose to run Rasa X.

Does this answer your question?