How to know a model in NLG / Action server

Rasa Open Source
1

Hello there, is it possible to pass an information about model into NLG and Action server requests?

Setup: We use our own NLG & Action server via URL

endpoints.yml

nlg:
    url: http://whatever.com/nlg

action_endpoint:
    url: http://whatever.com/action

And we want to be sure, which model is deployed to Rasa when it’s requesting NLG & Actions. So let’s say we call PUT /model to deploy a new model. It takes some time and immediately as it’s finished, NLG and Action server must be able to respect new deployed changes. Therefore we would need for example a header value from Rasa like

-> rasa calling http://whatever.com/nlg
headers: {
    X-MODEL=current_model_id
}

-> rasa calling http://whatever.com/action
headers: {
    X-MODEL=current_model_id
}

so that NLG and Action servers can serve a correct response.

Is this possible to achieve? We use Rasa 3.2.0

2

The status endpoint will return the loaded model_id and model_file.

3

Yes I know, but it seems to me a bit inefficient to call /status with every single NLG and Action server call