Running ML Agents inside Docker


I am having trouble launching an executable of a Unity ML Agents environment inside Docker. When I launch the executable inside the Docker with the following command, I receive a timeout error.

env = UnityEnvironment(file_name=env_name, worker_id=0, no_graphics=True, docker_training=True)
I tried with docker_training=True, and the same thing happens. I have mapped the base port of Unity ML Agents (5005) in the Docker container to a port on the host machine and have installed xvfb inside the Docker.

Is there anything else I am missing? The Unity executable is built on a Windows machine against a Linux platform. Is it possible that something is wrong with the executable?


1 Like

Hi @sahandrez , unfortunately we've deprecated our docker solution, and it probably doesn't work for now.