I have Rasa X 1.2.7 deployed and working using the official Helm chart.
I have also deployed two Rasa Opensource 3.4.0 stacks using the Rasa OS helm chart, one is serving the ‘production’ environment and the other the ‘devlocal’ environment. I have a valid model tagged with both environments.
When I share my assistant for guest testers for the production environment, I can chat with it normally in the chat ui. Not so for the devlocal environment.
In the devlocal chat, everything I type does arrive to the devlocal Rasa service and is processed normally. The bot also generates the correct response which is returned to the Chat UI as a result of the call to https://rasax-host/api/conversations//messages?environment=devlocal. But Chat UI never displays the response. I am assuming that the event itself never arrives through the web socket connection but I don’t know how to verify or debug this.
I have verified that the devlocal Rasa service does reach RabbitMQ and posts messages successfully.
What can be the culprit here?
An additional related question is: what queue should I configure for the various Rasa OS environments? I can configure different queues and they are created and filled, but only one queue (rasa_production_events by default) is consumed by the rasa X event service and I cannot configure it to consume multiple queues. Are all environments supposed to post to the same RabbitMQ queue, and if not, how are queues supposed to be handled?