Node operator guide

It stores and relays the network’s public records. It hosts the records the network assigns to it, relays other traffic, and can take on assigned cache work.

A Linux machine running Ubuntu 24.04 LTS, on x86-64 or ARM64, with a public address. The Bigfoot CLI that sets up and manages the node runs on Linux and macOS, on both ARM64 and x86-64. You do not need BFT to join: registering a node is free.

No. A storage node and a validator never share a machine or a public address. To run both, use two machines with two addresses.

Download the node.

No. Running unmodified node software, your node carries what the network assigns: you don’t choose, change, or remove records.

Modifying the software to select, change, or drop records turns you from a carrier into a curator: you’d be deciding what the network holds, and you’d be responsible for those decisions. The network also treats it as failing your assigned work, which costs standing and can lead to removal.

Operators are responsible for the laws that apply to them where they run a node. This guide is not legal advice.

Only what you actually have. Your node does not set disk space aside, and your offer is your commitment. If your node cannot store data the network assigns to it, the data goes to another node and yours loses standing. The network entrusts your node with more as its standing grows.

No. Relaying is part of running a storage node and cannot be switched off.

As available as you can make it. Your node builds standing through verified service, and unplanned downtime costs it quickly: each hour offline removes 14 hours of credited service. If a long absence uses up a node’s standing, the network starts removing it permanently.

Run bigfoot-node maintenance on, then bigfoot-node stop. maintenance on checks with the network that your node can stop safely and asks you to confirm, but it does not stop the node itself. Planned downtime is covered first by maintenance credit, which your node builds up at one hour for every 14 hours of verified service, up to 48 hours. New nodes start with none. When you are done, run bigfoot-node maintenance off, then bigfoot-node start.

Scheduled protocol upgrades install automatically at the height the network sets. bigfoot-node status shows any staged or scheduled upgrade, and every command prints a notice while a release is waiting for you. That release installs when you run bigfoot-node update.

BFT for assigned work the network confirms. Every new post and reply pays a service charge: the network credits 47.5 percent to the node that hosts it and 23.75 percent to each of the two nodes that relay it. Nodes are also credited for assigned cache work. Being online or offering capacity is not credited by itself, and carrying footnotes is not credited. Fees and credits lists every rate.

To the wallet you name as your payout owner, with bigfoot-node init --payout-owner THUMBPRINT or later with bigfoot-node configure-node --payout-owner THUMBPRINT.

Yes. Stop the node, back it up with bigfoot-node backup --out DIR, and restore it on the new machine with bigfoot-node restore --from DIR. Your node keeps its identity, standing, and duties. Copying the identity without its data does not work: the copy fails its duties and is banned.

Run bigfoot-node decommission and confirm. The node stops taking new work, the network moves the data it holds to other nodes, and when that is done the node’s identity and its local keys are permanently retired. Keep the node running until then. Decommissioning cannot be undone.

Your node’s history and its BFT are kept, and a permanent record of its retirement stays on the network. To run a node again, start a new one with a new identity.

The network treats it as an outage. Your node loses standing, and once its standing runs out it is removed permanently, with its unfinished duties still owed. Decommissioning is the clean way to leave.