RPC nodes are read-only Sigil nodes that serve public JSON-RPC and WebSocket traffic. They do not produce blocks. The same sigil-node binary runs in RPC mode under a different profile.

​
When to run one

  • Building a wallet, indexer, or dApp that needs predictable rate limits.
  • Operating a private cluster where validator-internal traffic must be isolated.
  • Serving a region not covered by public endpoints with adequate latency.

​
Hardware

ResourceMinimumRecommended
CPU8 cores16 cores
RAM32 GiB64 GiB
Storage1 TiB NVMe2 TiB NVMe
Network1 Gbps10 Gbps
RPC nodes do not need the validator’s HSM. Reads are signature-free.

​
Modes

ModeState retentionWhen to use
pruningLast 50,000 blocksWallet, dApp, exchange use.
archiveFull historyIndexers, explorers, analytics.
lightState at finalised blocks onlyEmbedded clients.

​
Configuration

/etc/sigil/rpc.toml: 
[node]
network = "sigil-mainnet-1"
data_dir = "/var/lib/sigil"
mode = "rpc"
pruning = "pruning"  # or "archive"

[rpc]
profile = "public"
bind = "0.0.0.0:8545"
max_request_bytes = 65536
max_send_tx_bytes = 4194304
max_batch_bytes = 16777216

[rpc.rate_limit]
rps = 25
burst = 100

[ws]
bind = "0.0.0.0:8546"
max_subscriptions_per_conn = 100

[metrics]
bind = "127.0.0.1:7778"

[p2p]
listen = "0.0.0.0:30303"
bootnodes = [
  "/ip4/65.21.46.10/tcp/30303/p2p/12D3KooW...",
]

​
Start

sigil-node start --config /etc/sigil/rpc.toml
Snapshot-restore is identical to the validator flow: 
sigil-node sync from-snapshot \
  --network sigil-mainnet-1 \
  --snapshot-url https://snapshots.sigil.ml/sigil-mainnet-1/latest.akasha.snap

​
Public-facing posture

If the RPC node is open to the Internet, front it with:
  • A reverse proxy (Caddy, Envoy) with TLS termination.
  • A rate-limiting layer (per-IP token bucket, per-API-key for premium tiers).
  • A WAF (Cloudflare or equivalent) — optional but recommended.
The Sigil public RPC fleet uses Cloudflare for the default endpoint and exposes a direct (no-middlebox) endpoint for censorship-resistance. See RPC endpoints for the production posture.

​
Health and metrics

  • GET /health returns 200 healthy / 503 unhealthy with the failed-probe list.
  • GET /metrics returns Prometheus-format metrics.
Critical alerts:
  • sigil_block_height lag against the public RPC for the network.
  • sigil_p2p_peers_connected < 8.
  • sigil_rpc_5xx_total rate.
  • Disk usage > 70%.
See monitoring for the recommended Grafana dashboards.

​
Premium tiers

To offer authenticated higher-rate tiers, deploy the auth sidecar: 
sigil-node auth-sidecar start --rpc-bind 0.0.0.0:8545
The sidecar issues JWTs after a challenge-response signing flow. See authentication for the protocol details.