Uptime > 2 epochs
Liveness failure (no participation in >2 consecutive epochs) costs 0.5% of stake and active-set removal. Re-registration goes to the back of the standby queue.
§ Developers · /developers/run-a-node
Run a node.
Four roles. Two staked (Node Runner, Node Watcher), two permissionless (Prover, Service Node). Any one of them can run on residential fibre. The reference client is a single Rust binary; configuration is a small TOML file.
§ 01 / Quickstart
Five commands.
# 1. Install (Rust 1.75+) cargo install --git https://github.com/adamant-protocol/adamant-node --tag v0.1.0 # 2. Initialise local data dir adamant-node init --role watcher --data-dir ~/.adamant # 3. Generate keys (Ed25519 + ML-DSA) adamant-node keys generate --out ~/.adamant/keys # 4. Edit config $EDITOR ~/.adamant/config.toml # 5. Run adamant-node run --data-dir ~/.adamant
§ 02 / Hardware
Per role.
| Role | CPU | RAM | Storage | Network | Stake |
|---|---|---|---|---|---|
| Node Runner | 4–8 cores | 16 GB | 2 TB NVMe | 1 Gbps symmetric, <100ms WAN | 1 000 ADM |
| Node Watcher | 2–4 cores | 8 GB | 500 GB SSD | 25 Mbps | 100 ADM |
| Service Node | 4+ cores | 16+ GB | 2+ TB | 1 Gbps | — |
| Prover | GPU recommended | 32+ GB | 200 GB | 100 Mbps | — |
§ 03 / Config
config.toml.
[node] role = "runner" # runner | watcher | service | prover data_dir = "/var/lib/adamant" chain_id = "0x4d41" # mainnet (adamant/mainnet) [network] listen_addr = "/ip4/0.0.0.0/tcp/30303" bootnodes = [ "/dns4/boot0.adamantprotocol.com/tcp/30303/p2p/12D3KooW…", "/dns4/boot1.adamantprotocol.com/tcp/30303/p2p/12D3KooW…", ] [rpc] listen_addr = "127.0.0.1:8545" ws_addr = "127.0.0.1:8546" [validator] bond = "1000_000_000_000" # 1 000 ADM in base units (10^-9) fee_payer = "self" [keys] ed25519 = "/etc/adamant/keys/ed25519.key" mldsa = "/etc/adamant/keys/mldsa.key"
§ 04 / Ops
Day-to-day.
Key rotation
Use a session subkey for the hot validator process; keep the master in cold storage. Rotate every 90 days as routine hygiene.
Snapshots
Reproducible snapshot every epoch; restore in <15 min from clean. Reference client ships adamant-node restore.
Monitoring
Prometheus metrics on /metrics. Default Grafana dashboards in the repo.
Backups
Recovery seed in two air-gapped locations. Slashing for double-sign is 100% — assume the worst.
Slot transfer
Transferable by mutual consent at an epoch boundary. The genesis cohort marker does not transfer with the slot.