Skip to main content

Using Docker

The Rasa Pro Docker image is available on Docker Hub as rasa/rasa-pro.

note

In the commands below, replace RASAVERSION with your desired version of Rasa Pro. To find the latest version of Rasa Pro, see the Changelog.

To pull the image, run:

docker pull rasa/rasa-pro:RASAVERSION

Optional Dependencies (Extras)

Rasa Pro supports optional dependencies through "extras":

  • nlu: Dependencies required for NLU components
  • channels: Dependencies required for channel connectors
  • full: Complete set of all optional dependencies

By default, the Rasa Pro Docker image already contains the nlu and channels dependencies.

Extending the Docker Image

To extend the Docker image with optional dependencies that are not already included in the default image, create a custom Dockerfile:

# Extend the official Rasa Pro image
FROM rasa/rasa-pro:RASAVERSION

# Install all optional dependencies
RUN pip install rasa[full]

# Alternatively, install individual optional packages rather than all optional dependencies comprised in `full`
RUN pip install rasa[spacy]

Build and run your custom image:

# Build the custom image
docker build -t my-custom-rasa:latest .

# Run the custom image
docker run -v ./:/app \
-e RASA_LICENSE=${RASA_LICENSE} \
-p 5005:5005 \
my-custom-rasa:latest \
run

Licensing

As explained in Licensing, at runtime you must provide the license key in the environment variable RASA_LICENSE.

After you defined the environment variable on your system, you can pass it into the docker container via the -e parameter, for example:

# execute `rasa init` via docker
docker run -v ./:/app \
-e RASA_LICENSE=${RASA_LICENSE} \
rasa/rasa-pro:RASAVERSION \
init --no-prompt --template tutorial