RasaX Chatbot Deployment

Run custom chatbot installation with Docker Compose

Minimum Requirements

• Ubuntu server 18.04+
• 2-6 vCPUs
• 4GB RAM
• 100GB Disk space
• Python 3.6 - 3.8 (for RasaX 0.42.0)

Setup

• Clone repo
• The default installation path is /etc/rasa , however, you can install in your desired path if you set: export RASA_HOME=~/rasa/dir
• Run the commands below to install

cd chatbot_deployment
sudo chmod 777 install.sh

# -E will preserve the environment variable you set
sudo -E bash ./install.sh

• cd into your installation folder (assuming you installed to /etc/rasa/)
• Run sudo docker-compose build to build custom images
• Run docker-compose build to build and image from rasa-x-custom
• Copy env-sample to .env file and update the environment variables
• Run sudo chown -R 1001:1001 db/ to change the permissions of the db postgresql local db folder
• Run docker-compose up -d to run the app stack
• Check running containers using docker ps

Once the app has started and database migrations complete, visit http://YOUR_IP_OR_DOMAIN on your browser.

To reset the default password, run sudo python rasa_x_commands.py create --update admin me YOUR_PASSWORD

Linking to Git

RasaX connects to a git repository where all training data is maintained. Use this guide to link to your repository.

Training in Container

You can train the bot inside the RasaX container as below:

• Access container shell: docker-compose exec rasa-x /bin/bash
• Train bot: rasa-train

Installing SSL certificate

We use Certbot as the provider for SSL certificates:

• Run ./install-ssh.sh to install certbot
• Update the domain in the script ./cp-certs.sh and run it to copy the certificates into the certs directory

Resources

• Install RasaX with Docker Compose
• Rasa Playground
• Securing with SSL