• Supervisor cli and ui to manage multiple rasa models
• Async HTTP API to put file and update model
• Train and update model without affecting current service¹

1 Switch between two instances

docker build . -t asar

docker run -d -p 5000:5000 -v ~/data:/data -v $(pwd):/app --name asar asar

• 5000: This server
• 9999: Supervisor
• /data: store model files to /data/model/name
• /app: auto reload code

docker logs -f --tail 1000 asar # server log, not supervisor log

install rasa and the required modules listed in Dockerfile, then run supervisord

Just see app.py, or http:https://localhost:5000/docs

• GET /supervisor supervisor ui
• GET /model/name get model info
• POST /model/name communicate to Rasa HTTP API
• PUT /model/name put files and update model

http:https://localhost:5000/supervisor

I'm not good at async, state, docker, supervisor, etc.

If you have any suggestion, please let me know.