Intention

This repository is intended to hold artifacts for creating a Flight Control (AKA Red Hat Edge Manager) homelab. If you want to learn more about this project, check out its repo here. Also included in this repository are commands and ContainerFiles for creating NVIDIA Jetson Red Hat Enterprise Linux Image Mode (bootc) images that will be used to flash edge devices managed by Flight Control.

Prerequisites

• A laptop running Red Hat Enterprise Linux 9.x
• DHCP server on your home network. This can be installed on the laptop if needed per these instructions or you can just let your home network router assign IP addresses.
• Ideally, use wired networking. Wireless networking will involve additional steps outside the scope of this repo.
• One or more NVIDIA Orin-based devices to manage (Nano, NX, AGX)

Initial Laptop Setup

Start with a server+GUI RHEL 9.x installation, with the "container management" software group selected. During RHEL installation, configure a regular user with sudo privileges on the host.

Make sure to set a hostname on the system other than localhost via:

sudo hostnamectl set-hostname <LAPTOP_HOSTNAME>

I have successfully deployed this setup on a NUC with 12 GB RAM and an Intel N97 processor, which are pretty minimal specifications, but you can likely go even lower on RAM if needed.

Once the system is back up:

• sudo dnf install -y git
• git clone https://github.com/tarexveff/jetson-rhem-workshop

Flight Control Server Installation

• Go into the ./jetson-rhem-workshop directory that you cloned
• Run: sudo ./1_configure-firewall-and-registry.sh
• Run: sudo ./2_rhem-server-build-commands.sh

Once these scripts are complete, you should be able to open the Flight Control web UI at https://LAPTOP_HOSTNAME. The login credentials are admin/admin unless you changed them in the 2_rhem-server-build-commands.sh script.

Post Installation Steps

With Flight Control now installed, you can generate the config.yaml file that you'll need to place on your managed devices to allow them to enroll to the server:

• flightctl login https://LAPTOP_HOSTNAME:3443 --insecure-skip-tls-verify --web
• flightctl certificate request --signer=enrollment --expiration=365d --output=embedded > config.yaml

Add the following line to the end of the config.yaml file and save it for use later!

• spec-fetch-interval: 5s

Managed Device Image Creation

Currently, building images on an x86 computer for NVIDIA ARM-based devices is very challenging, so I recommend creating an NVIDIA "build server." Bootstrapping this setup can be a challenge if you don't have an ARM server available, though you may be able to perform an initial build using an AWS Graviton instance if you don't have an NVIDIA system available.

Once you have a build server, you can use the resources in the managed-device-artifacts folder of this repository to make managed device images:

• jetson-bootc-build-commands: This contains the steps for creating a RHEL Image mode (bootc) ISO.
• ContainerFile-Jetson-rhem-1.0.2-base: This is the containerfile you will use to create the base image, which also includes the Flight Control agent (set to the same version as the server used in this repository).
• ContainerFile-Jetson-rhem-1.0.2-with-config: This builds on the image you created above, and adds the config.yaml file for your Flight Control server that you exported in the "Post Installation Steps."
• rhem.ks: This is a sample kickstart file for use in the mkksiso step of the jetson-bootc-build-commands build-commands.