ration/infra/dev-ca/README.md
caelum0x 13abc31645 feat(ration): observability, infra-as-code, security hardening, and test depth
Phase 2 sections 2-6 and 8 of docs/SCALE-EXPANSION.md:

- pkg/telemetry + pkg/metrics: OpenTelemetry trace propagation and
  Prometheus metrics across every service hop (CLI -> agent -> gateway ->
  card-issuance/tenant-registry/auth -> ledger-core), tracing added to
  ledger-core itself via services/ledger-core/src/telemetry.rs.
- infra/: otel-collector + Prometheus + Grafana docker-compose stack, a
  Terraform ECS Fargate module + dev/staging/prod envs (plan-only, no
  apply — real AWS spend is an owner action), and a dev-only mTLS CA
  (infra/dev-ca) wired through pkg/mtls + services/ledger-core/src/mtls.rs.
- services/audit-log (from the prior commit) gets its async mirror writer
  in services/ledger-core/src/audit_mirror.rs, and services/gateway gets
  a per-tenant token-bucket rate limiter (internal/ratelimit).
- Test depth: a Rust proptest suite proving concurrent ApplyEntry never
  breaches the reserve floor (tests/apply_entry_proptest.rs, passing), a Go
  fuzz seed corpus for the policy envelope (rules_fuzz_test.go), a libFuzzer
  harness for the C WAL frame parser (services/wal/fuzz, deliberately not
  wired into `make test` per its own Makefile comment - needs a
  libFuzzer-bundled clang), and a real chaos test
  (scripts/chaos/kill-ledger-mid-write.sh) that SIGKILLs ledger-core
  mid-write under concurrent load and verifies via ListEntries that
  nothing was lost or double-applied.
- docs/formal/ledger.tla: a TLA+ spec of the ledger state machine (balance
  >= reserve_floor safety, HOLD-eventually-resolves liveness) plus a TLC
  model config - not run through TLC in this pass, that's a precise
  documented TODO, not faked.
- docs/adr/0001-0004, docs/RUNBOOK.md, docs/SLO.md, docs/THREAT-MODEL.md.

Verified: `make build && make test` clean (C WAL, Rust incl. the new
proptest suite, all 10 Go modules) from a cold target.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-10-19 12:25:58 +03:00

2.8 KiB

infra/dev-ca — dev-only internal CA

Generates a throwaway certificate authority and one cert+key pair per Ration service, so mTLS between internal services (pkg/mtls on the Go side, services/ledger-core/src/mtls.rs on the Rust side) can be exercised locally and in the demo without any real PKI.

Dev-only — do not use any of this in a real deployment. The CA private key lands on disk in plaintext right next to everything it signs, certs live for a year with no rotation story, and there is no revocation. A real deployment should issue short-lived service identities from a managed CA instead — AWS Private CA / ACM PCA, HashiCorp Vault's PKI secrets engine, or a service mesh's built-in mTLS (Istio/Linkerd) — none of which this script tries to be.

Usage

infra/dev-ca/generate-dev-ca.sh            # writes to infra/dev-ca/out/
infra/dev-ca/generate-dev-ca.sh /some/dir  # or a custom output dir

Output (out/ is gitignored — regenerate it, don't commit it):

File What
ca.pem / ca.key The dev CA's own self-signed cert/key.
<service>-cert.pem / <service>-key.pem One identity per service (ledger-core, gateway, audit-log, agent, cli, tui), signed by ca.pem. Each cert carries serverAuth and clientAuth EKUs and SANs for <service>, localhost, 127.0.0.1, since every internal service here both serves its own RPCs and dials others.
dev-ca.env export FOO_MTLS_ENABLED=true / _CERT_FILE / _KEY_FILE / _CA_FILE for every service, matching the env var convention pkg/mtls.LoadFromEnv(prefix) (Go) and ledger-core's Rust loader read.

To actually require mTLS for a local run:

set -a
source infra/dev-ca/out/dev-ca.env
set +a
make build
./bin/ledger-core &          # now requires + verifies a client cert
./bin/gateway &               # dials ledger-core with its own cert,
                               # and requires one from its own callers

Leave the env vars unset (the default) and everything runs exactly as it did before this existed — plaintext gRPC, no certs required. Every service in this repo treats mTLS as opt-in, matching the AIRWALLEX_MODE=mock|live pattern already used for the Airwallex integration.

Why one identity reused as both server and client cert

Ration's internal services form a small, symmetric mesh (gateway dials ledger-core, ledger-core best-effort-mirrors into audit-log, the TUI/CLI dial gateway directly) rather than a strict client/server hierarchy, so issuing a single cert per service with both EKU bits set is simpler to operate than maintaining separate server and client identities per service — at dev-CA trust-boundary granularity (see docs/THREAT-MODEL.md) this distinction doesn't buy anything a real deployment's managed CA wouldn't already handle through separate short-lived certs per workload identity anyway.