Running `alpyperl` in a docker container

If you are planning to develop your project in a docker container, you must be aware of the following:

Connection between python and AnyLogic will potentially fail. This is due to alpyperl trying to connect to a port that is not mapped between the docker container and docker host.

Howerver, this is possible when running with an exported model as execution of the model will take place in the container. In the terminal, AnyLogic will share the port in case you want to access the model UI.

root@e71cad1ec378:/workspaces/ALPypeRL# python alpyperl/examples/cartpole_v0/train.py
2023-03-14 22:10:20,619 INFO worker.py:1553 -- Started a local Ray instance.
(RolloutWorker pid=3831) /usr/local/lib/python3.10/site-packages/gymnasium/spaces/box.py:127: UserWarning: WARN: Box bound precision lowered by casting to float32
(RolloutWorker pid=3831)   logger.warn(f"Box bound precision lowered by casting to {self.dtype}")
(RolloutWorker pid=3831) chmod: cannot access 'chromium/chromium-linux64/chrome': No such file or directory
(RolloutWorker pid=3831) Couldn't find browser: chromium/chromium-linux64/chrome. Attempting to open system-default browser for url: http://localhost:23109
(RolloutWorker pid=3831) [INFO] ALPypeRLConnector - Welcome to the 'ALPypeRL Connector'!

If you are planning to host your trained policy in a docker container you should apply the same logic: you must map the port to the host machine so it is accessible. Then, just remember to set the right port in the ALPypeRLConnector.

Running alpyperl in a docker container

Running `alpyperl` in a docker container