Production hardening checklist

A pragmatic checklist for a production deployment. Many items are already the chart’s defaults — this is what to confirm.

Identity & secrets

• Auth enabled (AUTH_ENABLED=true) with at least one key; keys stored hashed (sha256:<hex>).
• Rate limiting enabled, Redis-backed (shared across instances).
• Fabric signing identity is least-privilege (own bundle, key 0400), not the whole org tree.
• Secrets in a secrets manager, not in ConfigMaps or images; rotation planned.

Data & durability

• WAL on durable storage (PVC), not an emptyDir.
• External managed MongoDB (replica set + backups); mongodb.enabled=false.
• Backup & restore tested — and verified against on-chain roots (see backup & DR).

Schema & chaincode

• Migrations run by the pre-upgrade Job, not the API entrypoint; the API asserts the version.
• Chaincode upgrade path scripted and rehearsed (chaincode lifecycle).

Isolation

• Per-tenant identities provisioned and mapped (fabric.tenant_identities) if you run multiple tenants — not just API-key isolation.
• Per-tenant channels considered for the strongest ledger isolation.

Network & surface

• API behind TLS / an ingress with your own hostname.
• CORS scoped to known origins (not *) if browsers call it with credentials.
• Request body cap and timeouts set for your workload.

Observability

• /health wired to liveness/readiness; Prometheus scraping :9090/metrics.
• Alerts on anchor failures, batch backlog, error rate, Fabric health.
• make smoke (or equivalent) on a schedule.

Scale

• Single API replica or per-pod WAL persistence arranged before setting allowMultiReplica=true.