Ml model delivery to docker container

I want to up container with web server which use knn model. But i need to retrain knn model every quartal. What is the best way to do it?

1. make post query to web server on special url which will start retrain model directly in container
2. retrain model outside of container and use docker volumes to deliver new version

As for the first approch, how to limit other users, only I can write requests to this url.