How to add `remote-storage` in rasa-x deployment?

According to the Rasa Open Source documentation you can add --remote-storage aws to the rasa run command (with all the needed environment variables) to load models from s3.
Since the deployment of rasa-x is slightly different, I’m not able to load models from s3. To what docker compose service do I have to add the --remote-storage flag and how to add it to this service (rasa-x / rasa-worker / rasa-production)?

When I add the flag to rasa-worker & rasa-production it says the following (don’t mind the 127.0.0.1 part):

Model server could not find a model at the requested endpoint 'http://127.0.0.1:5002/api/projects/default/models/tags/production'. It's possible that no model has been trained, or that the requested tag hasn't been assigned.

@koaning @b-quachtran is this anything you can help with?

@sschrijver Unfortunately Rasa X doesn’t support the remote cloud storage options that are available in Rasa Open Source.

To externalize models with Rasa X, you’ll have to create a model server and point Rasa X to that endpoint. There’s some documentation on how to do that here: Model Storage

Hi @b-quachtran,

Thanks for your reply. Are there any guides/best practices to setup a model server?