Rasa Deployment in AWS cluster mode

[Nagasuman (Suman)]

Hi Team,

I am working on the RASA Chatbot in my organisation. They want to use AWS cluster deployment for the same. We need to know how to configure RASA open source service within EC2 instance (In multiple nodes) and that will share a Redis DB from another instance for session tracking. By doing this if one of the RASA service is going down then another instance should take over the conversation with the help of Redis tracker.

Similar requirement is posted here:

How to configure rasa tracker store without docker

Hello! I’ve been asked to setup Rasa open source as a service with a configured tracker store (Redis, preferably). The idea is to have multiple nodes running Rasa NLU and Core with a shared tracker store to avoid SPOF. In the event of a node going down, the standby node is expected to take over the conversation. From my understanding, a tracker store would serve the purpose. However, the proposed set up is not a docker. Was curious to know if this is possible because I did not come across any …

Thanks for the help!!

Thanks, Suman.

[degiz (Alexander Khizov)]

Hey @Nagasuman

We need to know how to configure RASA open source service within EC2 instance

I don’t see a huge difference between AWS EC2 and any other VPS, so we usually recommend to follow steps described in our docs.

Hope that helps!