Steve Pham 76b40e6f34 Revert "Fix permission issues in docker container for the installed o3de (#18047)" (#18055) 5 maanden geleden
..
Dockerfile 5336271b9c O3DE Docker support (#17998) 5 maanden geleden
README.md 5336271b9c O3DE Docker support (#17998) 5 maanden geleden
build.sh 5336271b9c O3DE Docker support (#17998) 5 maanden geleden
entrypoint.sh 76b40e6f34 Revert "Fix permission issues in docker container for the installed o3de (#18047)" (#18055) 5 maanden geleden

README.md

Docker support for O3DE

O3DE supports construction of a Docker image on the Linux environment. The following instructions will allow you to build and launch O3DE through a Docker container, with project data being persisted on your local machine.

Prerequisites

  • Hardware requirements of o3de
  • Any Linux distribution that supports Docker and the NVIDIA container toolkit (see below)
  • At least 60 GB of free disk space
  • Docker installed and configured
    • Note It is recommended to have Docker installed correctly and in a secure manner so that the Docker commands in this guide do not require elevated priviledges (sudo) in order to run them. See Docker Engine post-installation steps for more details.
  • NVidia container toolkit

Building the Docker image

The Docker script provides the following arguments for building the Docker image, with each argument provide a default value.

Argument Description Default
INPUT_IMAGE The base ubuntu docker image to base the build on ubuntu
INPUT_TAG The base ubuntu docker image tag to base the build on jammy
INPUT_ARCHITECTURE The CPU architecture (amd64/aarch64). Will require QEMU if cross compiling amd64
O3DE_REPO The git repo for O3DE https://github.com/o3de/o3de
O3DE_BRANCH The branch for O3DE development
O3DE_COMMIT The commit on the branch for O3DE (or HEAD) HEAD

Examples

The following example build commands are based on the general use case for O3DE.

If there are any ROS2 gem based projects that will be created or used by the image, then you will need to follow the ROS Example section below.

You can build a minimal Docker image that can create or use non-simulation projects using the ubuntu input image and either the jammy or noble input tag. To build a Docker image that supports robotic simulations using ROS, you must use the ros input image and either the humble or jazzy input tag. (Refer to the Docker distribution of ROS)

Locally

To build the Docker image from the local O3DE Docker context, run the following command from the Docker subfolder inside the o3de root folder.

General

docker build -f Dockerfile --build-arg INPUT_IMAGE=ubuntu --build-arg INPUT_TAG=jammy -t amd64/o3de:ubuntu.jammy .

ROS2 enabled

docker build -f Dockerfile --build-arg INPUT_IMAGE=ros --build-arg INPUT_TAG=humble -t amd64/o3de:ros.humble .

From github

You can also build the Docker image from the O3DE Docker context directly from github with the following command (based on the development branch)

General

docker build -t amd64/o3de:ubuntu.jammy https://github.com/o3de/o3de.git#development:Docker

ROS2 enabled

docker build --build-arg INPUT_IMAGE=ros --build-arg INPUT_TAG=humble -t amd64/o3de:ros.humble https://github.com/o3de/o3de.git#development:Docker

Running the Docker image locally

Docker containers by themselves are stateless, so in order for the container to act as a stateful container, some local preparations and additional runtime arguments are needed to be passed into the Docker run command.

Setting up O3DE-specific folders

O3DE will use two folders for its runtime environment. The manifest folder (\$HOME/.o3de) and the O3DE Home folder (\$HOME/O3DE). On normal installations/build of the full O3DE engine, these folders are created on-demand based on the current logged in user. Since Docker containers by default are stateless, all the information that is downloaded, updated, deleted, etc from these two folders as part of normal operations will no longer exist once the container is disconnected, thus losing any work that was done while using O3DE. In order to utilize the O3DE Docker container to create and maintain O3DE projects beyond the container session, we will need to create these two folders manually and pass them in as mapped volumes to its corresponding specific folders within the container.

The following example creates these folders under a parent docker folder in the current user's $HOME directory, specific to the name of the Docker image.

mkdir -p $HOME/docker/o3de/manifest

mkdir -p $HOME/docker/o3de/home

Note The two folders must be created specifically for the O3DE Docker container that is being connected to. There is no version or compatibility checks in the Docker container, so normal O3DE startup in the container may fail if it does not use the same mapped folders it used during its initial launch.

These two folders will be added as part of the docker run argument list:

-v "$HOME/docker/o3de/manifest:/home/o3de/.o3de" -v "$HOME/docker/o3de/home:/home/o3de/O3DE"

Using the current user id and group id

The Docker image will map the current user and group id to a Docker-created user and group when running the container. This will allow reading and writing to the mapped O3DE-specific folders described earlier and will add persistence to the Docker container through the mapped folders. The user id and group id are passed to the Docker run command as two arguments: UID and GID:

--env UID=$(id -u) --env GID=$(id -g)

Putting it all together, you can launch the O3DE Docker container with the following command:

xhost +local:root

docker run --env UID=$(id -u) --env GID=$(id -g) -v "$HOME/docker/o3de/manifest:/home/o3de/.o3de" -v "$HOME/docker/o3de/home:/home/o3de/O3DE" --rm --gpus all -e DISPLAY=:1 -v /tmp/.X11-unix:/tmp/.X11-unix -it amd64/o3de:ubuntu.jammy

For the ROS container based example, add the ros.humble tag.

xhost +local:root

docker run --env UID=$(id -u) --env GID=$(id -g) -v "$HOME/docker/o3de/manifest:/home/o3de/.o3de" -v "$HOME/docker/o3de/home:/home/o3de/O3DE" --rm --gpus all -e DISPLAY=:1 -v /tmp/.X11-unix:/tmp/.X11-unix -it amd64/o3de:ros.humble

Once the Docker container is launched, you will be logged into the O3DE-enabled terminal. At this point, typing in o3de in the command line will launch the O3DE Project Manager where you will be able to create, build, and launch O3DE projects.