Hi, My bot is ready to put on production. I’m using only Rasa (NLU and Core). I have two python servers that both answer to a VIP (balancer). So, each one will respond to a request but, right now, I can’t choose which one. I would like to know if its possible do syncronize both servers, so when a request that is part of a conversation can be processed as a part. If it’s not possible, my chatbot may not work as it should. Does anyone know if it’s possible? I looked at docs and always show a single machine. (I have on-premise windows servers) thanks Kassim

I am not sure I fully understand what you mean. So you want the conversations to be sticky and always answered by the same instance? If so, we are currently working on making that easier. In the meantime that is something the load balancer needs to do (e.g. IP based, header based, content based).

Deployment on balanced server

Rasa Open Source

kassimorra (Kassimorra) June 25, 2019, 1:37pm 3

Maybe. Is there any way to keep two instances synced ? Then, I won’t have any trouble.

Topic		Replies	Views
Add conversation persistence with load balancer (nginx, redis, found possible solution) Rasa Open Source	1	658	July 29, 2022
How to have a load balancer on top of multiple rasa open source assistants Rasa Open Source	2	553	May 7, 2021
What is the recommended approach to send External Events to a load balanced Core Server stack Rasa Open Source	5	776	July 1, 2020
Recovery and Scalability for Rasa Bot Service(s) Rasa Open Source	8	2211	March 19, 2021
Have Multiple Agents on a Web Server Rasa Open Source	8	2643	September 5, 2018

Deployment on balanced server

Related topics