ankaios

Ankaios Deployment Guide

This guide explains how to deploy our system by using Ankaios

Overview

The ankaios deployment configulation is designed to:

1. Deploy Kuksa Data Broker on a Raspberry Pi acting as both the Ankaios server and Agent A.
2. Deploy G29 Publisher on another Raspberry Pi as Agent B.

The Kuksa Data Broker listens on 192.168.1.99:55555. You can connect any Kuksa client to interact with the broker.

Prepare the environment:

• Connect 2 Raspberry Pis to the same network.
  • OS used were Ubuntu 24.04.1 and raspberrypi 6.6.51.
  • IP of raspberry pi for server should be 192.168.1.99
• install ankaos v0.5.0
• install podman v3.4.4

please see : https://eclipse-ankaios.github.io/ankaios/latest/usage/installation/

Folder Structure

.
├── README.md
├── deployment.yaml  # Deployment configuration file 
├── startup-agent-a.sh # Script to set up Raspberry Pi as Ankaios server and Agent A  # Script to set up another Raspberry Pi as Ankaios Agent B
├── startup-agent-b.sh
├── test # Simple deployment configuration files to test(& learn) Ankaios behavior
└── val.json # custom VSS definition for kuksa broker

Deployment Steps

1. Set up the Ankaios server and Agent A:

   • Git clone and move to challegers/ankaios on the first Raspberry Pi.
   • run the following command:

     ./startup-agent-a.sh

   • This script configures the device as the Ankaios server and Agent A.

2. Set up Ankaios Agent B:

   • Git clone and move to challegers/ankaios on the second Raspberry Pi.
   • run the following command:

     ./startup-agent-b.sh

   • This script configures the device as Ankaios Agent B.

3. Check the connected agents

   • run the following command on one of Raspberry Pi.

   ank -k get agents
   

   It should print:

   NAME           WORKLOADS      CPU USAGE      FREE MEMORY
   agent_A        2              42%            42B
   agent-B        0              42%            42B
   

4. Deploy

   • run the following command on one of Raspberry Pi.

   ank -k apply deployment.yaml
   

5. Check the deployment

   ank -k get workloads
   

   It should print:

   WORKLOAD NAME   AGENT     RUNTIME   EXECUTION STATE          ADDITIONAL INFO
   databroker      agent_A   podman    Running(Ok)
   g29-publisher   agent_B   podman    Running(Ok)
   

Containerizing Applications

both application (Kuksa Data Broker and G29 Publisher) are already containerized and placed in a container repository.

Kuksa Data Broker

• github: https://github.com/eclipse-kuksa/kuksa-databroker
• container: ghcr.io/eclipse/kuksa.val/databroker:0.4.1

G29 Publisher

• github: https://github.com/Eclipse-SDV-Hackathon-Chapter-Two/challengers/tree/main/G29-publisher
• container: docker.io/bleach31/g29-publisher:latest