Autonomys - Ubuntu CLI - Cluster Guide
Introduction
This guide covers how to get everything set up to farm on the Autonomys network with clusters using the CLI in Ubuntu 24.04. Clusters are beneficial if you want to share plotting and farming resources across hosts. A cluster scales very well, and redundancy can be added to ensure your systems remain highly available and resilient against failures.
This guide should work just fine for Ubuntu Desktop or Ubuntu Server. I only use the CLI and prefer to SSH into my PCs. If you would like to become more familiar with the CLI this is a great guide to help get used to running commands in the CLI. I use Termius (free) for my SSH manager, and more information can be found in the SSH Manager Guide
How this guide works is I have a bunch of wikis which cover a specific step of the guide. I will give a quick intro for each wiki, and then provide a link. You should be able to just follow the wiki and move to the next step after each one.
Resources
Forward Ports
You must forward the ports that the Autonomys Node will use. This is port 30333 and 30433. To learn more about forwarding ports, check out this guide:
Nvidia Drivers (GPU Plotting Only)
Plotting speed can be greatly increased by a GPU. However, Ubuntu must have at least Nvidia drivers with a version of 550 or higher. Drivers below this version should be updated to the latest
Prepare Farm Disks
In order to Farm on Autonomys, you must dedicate SSD space which will have data plotted to it and then farmed for rewards. The more space you dedicate the more rewards you will earn once that space is plotted. Plotting can be pretty quick, with modern RTX 3000 and 4000 cards capable of plotting many Terabytes per day.
This guide assumes you will be dedicating an entire SSD to Autonomys. The following guide will walk you through how to wipe, format, mount, and permission your SSD for the Autonomys network.
Launch Node
Now that the disks are prepped its time to launch your Node. There are two parts to Farming on the Autonomys network: Node & Farmer. The Node connects to the Blockchain and the Farmer connects to the Node. They work together to participate in the Autonomys network. The Node will need to sync with the blockchain but it is usually pretty quick.
NATS as a Service
Before starting your Cluster you need to have NATS running. This can be run as a service using the following guide
Launch Cluster
Now that the Node is running, it is time to launch the cluster so you can begin plotting and farming your drives. This guide is specifically for cluster setups. What this means is that the Farmer is split into 4 applications: Controller, Cache, Plotter, Farmer.
Grafana (Optional)
Grafana allows metrics exposed via Prometheus to be used to generate robust dashboards that help you keep track of various metrics related to your Node, Farmer, System, and GPU. This step is optional but HIGHLY recommended. Grafana and Prometheus are both free. First, Grafana will need to be installed. After that, Prometheus will need to be installed. Then for each type of metrics you want to create a dashboard for, those metrics will need to be exposed to Prometheus. This and the following guides will walk you through it.
Prometheus Monitoring (Optional)
The Node and the Farmer both publish metrics that can be consumed by Prometheus. To access these metrics Prometheus can be run as a service on your PC. Install Prometheus as a service
Now the Autonomys jobs need to be added to the Prometheus config.
Node Exporter (Optional)
I like to keep an eye on my system resources such as RAM, CPU, Disk, and Network usage. If already planning to use Prometheus to monitor your Node and Farmer, you can install a service for node_exporter which will expose system resource information to Prometheus. This will allow you to create a dashboard to monitor your system resources in Grafana.
Nvidia GPU Exporter (Optional)
Unfortunately Node Exporter does not offer GPU monitoring. If you will be plotting via your GPU you may want to monitor the power, memory and usage. The Nvidia GPU Exporter does just that.