documentation:containers [robotwiki]

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== Containers ======

===== Prerequisites =====

  * Ubuntu 18.04, 20.04, 22.04
  * Feodora 34+
  * For rootless use, Podman must be preinstalled
  * You need ''git'' and ''make'' commands installed for ROS containers to work.
  * //Optional but useful:// [[https://docs.docker.com/engine/install/|Docker]] (can be installed for Debian/Ubuntu like dists with ''make install-docker'' in ROS containers)
  * //Optional but useful:// Nvidia Docker if you have a Nvidia GPU: https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html

===== SSH Keys =====
It can be configured in a number of ways, below is a suggestion:

Edit or create ''~/.ssh/config'':

<code>
Host git.cs.lth.se
    Hostname git.cs.lth.se
    User git
    IdentityFile ~/.ssh/id_gitcs
    IdentitiesOnly yes
</code>

Generate your key:

<cli>
ssh-keygen -t ed25519 -f ~/.ssh/id_gitcs
</cli>

You need to add your key (''~/.ssh/id_gitcs.pub'') to:
https://git.cs.lth.se/-/profile/keys

If you use a passphrase (simplifies access and the passphrase is stored in you keychain):
<cli>
ssh-add ~/.ssh/id_gitcs
</cli>

===== ROS Containers Quickstart =====

Valuable information about how to use ROS Containers exists in the repository readme:
https://git.cs.lth.se/robotlab/ros-containers 

==== Clone the ROS-containers repository ====
<cli>
git clone git@git.cs.lth.se:robotlab/ros-containers.git
cd ros-containers
</cli>

==== Set options ====
There are many options, all start with ''use-''. All use commands modifies ''config.env'' which you can modify in your favorite text editor aswell. Initially no ''config.env'' exists, it will be copied from ''config.env.template''.

Quickstart for ROS Noetic and Docker:
<cli>
make use-docker use-noetic
</cli>

Quickstart for ROS Noetic and Nvidia docker (for GPU compute and acceleration):
<cli>
make use-nvidia-docker use-noetic
</cli>

Both above requires your user to have ''sudo'' access. If your user belongs to the group ''docker'', you can instead use ''use-docker-rootless'' and ''use-nvidia-docker-rootless''. Adding your user can normally be done with the command ''sudo usermod -a -G docker [your usernane]''.

If you intend to run this with a real robot:
<cli>
make use-real
</cli>
This adds ''--priviledged'' flag to when running the docker image. This means that it would then no longer be fully sandboxed. This allows among other things unrestricted access to USB devices which is commonly required for cameras.

More options are available in the documentation: https://git.cs.lth.se/robotlab/ros-containers and more specifically you can take a peep inside ''Makefile'' to see available commands.

==== Build and run ====
<cli>
make run
</cli>

This will build the base image, a special user image where your user is known and mapped and finally a container to run it all in.

By default your home directory is mapped to ''home'' relative to ros-container repo root.

==== Resume and reset ====
Delete your container to change mappings or otherwise clean out your environment:
<cli>
make delete
</cli>

If you exit your container/restart your computer:
<cli>
make resume
</cli>

This command will resume your container.
// Note:// if the container is resumed in one terminal and you resume again, you will effectly use the same terminal interface in multiple terminals (they will be mirrored).


===== Quickstart: Heron =====
You should have succeeded in building the container and is inside it for below to work.

==== Build the Heron Workspace ====
<cli>
workspace init heron.rosinstall
workspace install
</cli>

==== Run Heron simulation ====
Two terminals or tmux/start roscore in background:

**Terminal 1:**
<cli>
workspace shell
mon launch heron_launch simulation.launch
</cli>

A note of behaviour, if you exit the first terminal you created with ''make run'' or ''make resume'' all others will be terminated  with the exit of the first.

**Terminal 2:**
New terminal running inside the container can be created with ''make terminal''
<cli>
roscore
</cli>


===== Workspace Tips/Tricks =====

==== Using a custom rosinstall ====
Let say it is named ''custom.rosinstall'', replace the name with whatever you want.

  - Copy your ''custom.rosintall'' to ''[ros-containers-path]/home'' or if using //shared-home//, to your user home directory.
  - In your home directory, run ''workspace init custom.rosinstall''
  - This will clone all the repos described in ''custom.rosinstall''

**Note:** ''workspace init'' autocompletes when pressing the [TAB] key, it combines central rosinstalls with local ones in your current directory.

==== Switching branches ====
By default ''~/catkin_ws'' will contain a file ''current.rosinstall'' which was used to initialize the workspace.
If you modify it, you can run ''workspace reimport'' to apply the new branches and clone new repos that did not exist before.

If you are using ROS2, it would be called ''~/colcon_ws'' instead.

**Note**: ''workspace reimport'' will not move or remove repos - you will have to do that manually.
====== Resources ======
===== WSL 2 GPU Drivers =====
https://learn.microsoft.com/en-us/windows/wsl/tutorials/gui-apps

===== VS Code Workflow =====
The files in the home folder (in the container) are by default available in ''home''. But instead of opening that folder in your VS Code, a more convenient way is to attach VS Code to the container. You need to have the [[https://marketplace.visualstudio.com/items?itemName=ms-vscode-remote.vscode-remote-extensionpack|Remote Development extension]] installed in VS Code for this to work.

To attach your VS Code to the running Docker container, choose the Docker panel and right-click on the container to get a context menu.

{{:documentation:vscode_attach_docker.png?400|}}