Deployment patterns

The reference node is a single Go binary. It deploys like any stateful service. Quidnug ships first-party manifests for the common topologies.

Single-node (development)

./bin/quidnug
# Listening on :8080, state in ./data/

Suitable for local development, demos, and early integration testing. No peer configuration required.

Three-node consortium (pre-production)

deploy/compose/ ships a Docker Compose stack with three nodes + IPFS + Prometheus + Grafana. Bring it up with:

cd deploy/compose/
docker compose up

The three nodes seed each other via SEED_NODES, share trust over gossip, and each produces its own Proof-of-Trust tiered chain.

Kubernetes (production)

deploy/helm/quidnug/ is a production-grade StatefulSet chart with:

PersistentVolumeClaims for each replica, the nonce ledger, trust registry, and block store are durable per-pod.
PodDisruptionBudget for graceful rollouts.
Anti-affinity so replicas land on different nodes.
HPA stub, horizontal scaling is per-role, not per-shard (the protocol is not sharded).

helm install quidnug deploy/helm/quidnug/ \
  --namespace quidnug --create-namespace \
  --values my-values.yaml

TLS termination

Terminate TLS at your ingress (Cloudflare, nginx, Envoy, Caddy, AWS ALB). The node itself speaks plain HTTP on the bound port.

This is the recommended topology: your ingress already handles certs, rate limiting, and WAF. Don’t re-implement inside the binary.

Inter-node HMAC authentication

Set NODE_AUTH_SECRET=<32-byte-hex> and REQUIRE_NODE_AUTH=true on every node in the consortium. Gossip, probe, and snapshot endpoints then require a valid HMAC over the request. External (client) endpoints continue to authenticate via the signed transaction itself, the HMAC gates only node-to-node traffic.

Inter-region deployment

Use the standard seeding mechanism; peers discover each other asynchronously.
Prefer a small core set of operators with direct mutual trust edges (high-weight); let broader membership attach via those.
Set a sensible RATE_LIMIT_PER_MINUTE per node so a misbehaving peer cannot cause resource exhaustion.

Upgrading

Use QDP-0009: Fork-Block Migration Trigger to coordinate protocol-level upgrades across a federated operator set. Binary upgrades (non-protocol) are rolling.