Day 1
Task Types
There are two primary ways we will conduct tasks. Both of the tasks must be performed from the shared user account k3s.
1. Group Task
What it is: A task performed by the entire team together in a single, shared terminal session.
Objective: To collaborate on configurations, provide training, or perform changes that require peer review.
Method: We will use the
GNU Screenprocedure.
Group Sessions
Switch to the shared user. You will be prompted for the shared user's password.
sudo su - k3sJoin the shared session.
screen -x group_sessionTo exit a screen session:
2. Separate Task
What it is: A task performed by an individual working alone in their own private terminal session.
Objective: For routine duties, research, or work that does not require collaboration.
Method: A standard, individual SSH connection, using the shared user account
k3s.
SSH into the required node.
Device Types
Our cluster is composed of several different types of hardware. Here is a brief overview of each device's role and characteristics.
control_plane
This device is an Amazon Web Services (AWS) EC2 virtual server instance. The specifications for this instance are detailed in the table below.
Specification | Value |
|---|---|
Instance Type |
|
vCPUs | 2 |
OS Image | Ubuntu 24.04 LTS (ARM64) |
**Location ** | Cloud (AWS) |
AGX
The AGX Device is an on-premise, high-performance computing unit. The hardware is an NVIDIA Jetson AGX Orin module, which is specifically designed for demanding AI and machine learning workloads at the edge.
Key specifications are listed in the table below.
Specification | Value |
|---|---|
Hardware Model | NVIDIA Jetson AGX Orin (64GB Module) |
CPU | 12-core Arm® Cortex®-A78AE v8.2 64-bit |
GPU | 2048-core NVIDIA Ampere architecture |
AI Performance | 275 TOPS (INT8) |
Memory | 64 GB 256-bit LPDDR5 |
Storage | 64GB eMMC 5.1 |
Networking | 10GbE Ethernet Port |
Location | On-Premise |
Nano
The Nano Device is a compact, on-premise computing unit designed for entry-level AI applications at the edge. The hardware is an NVIDIA Jetson Orin Nano Developer Kit, making it ideal for prototyping, robotics, and various educational projects.
Key specifications are listed in the table below.
Specification | Value |
|---|---|
Hardware Model | NVIDIA Jetson Orin Nano (8GB Dev Kit) |
CPU | 6-core Arm® Cortex®-A78AE v8.2 64-bit |
GPU | 1024-core NVIDIA Ampere architecture |
AI Performance | 40 TOPS (INT8) |
Memory | 8 GB 128-bit LPDDR5 |
Storage | microSD card slot (+ NVMe via M.2 Key M) |
Networking | Gigabit Ethernet Port |
Location | On-Premise |
Pi
The Pi Device is a versatile, low-cost, on-premise single-board computer. Based on the Raspberry Pi 5, this hardware is used for a wide range of general-purpose and lightweight tasks, from learning and experimentation to hosting simple applications and collecting sensor data.
Key specifications are listed in the table below.
Specification | Value |
|---|---|
Hardware Model | Raspberry Pi 5 (8GB) |
CPU | Quad-core Arm® Cortex-A76 @ 2.4GHz |
GPU | VideoCore VII GPU |
Memory | 8GB LPDDR4X SDRAM |
Storage | microSD card slot (+ PCIe 2.0 interface) |
Networking | Gigabit Ethernet Port |
Wireless | Dual-band 802.11ac Wi-Fi / Bluetooth 5.0 |
Location | On-Premise |