Merge pull request #4 from payneio/feat/mesh-tls

Mesh TLS hardening (NATS + OpenBao)

castle-api/src/castle_api/config.py

@@ -17,6 +17,7 @@ class Settings(BaseSettings):
     # Mesh coordination (all off by default — single-node works without them)
     nats_enabled: bool = False
     nats_url: str = "nats://localhost:4222"
+    nats_token: str | None = None
     mdns_enabled: bool = False
 
     model_config = {

castle-api/src/castle_api/main.py

@@ -68,6 +68,7 @@ async def lifespan(app: FastAPI) -> AsyncGenerator[None, None]:
                 local_hostname=registry.node.hostname,
                 local_registry=registry,
                 servers=settings.nats_url,
+                token=settings.nats_token,
             )
             await nats_client.start()
             app.state.nats_client = nats_client

castle-api/src/castle_api/nats_client.py

@@ -50,10 +50,12 @@ class CastleNATSClient:
         local_hostname: str,
         local_registry: NodeRegistry,
         servers: str | list[str] = "nats://localhost:4222",
+        token: str | None = None,
     ) -> None:
         self._local_hostname = local_hostname
         self._local_registry = local_registry
         self._servers = servers
+        self._token = token or None
         self._nc: nats.NATS | None = None
         self._kv = None
         self._presence_kv = None
@@ -77,9 +79,13 @@ class CastleNATSClient:
 
     async def start(self) -> None:
         """Connect, publish our registry, seed state, and start watchers."""
+        # A `tls://` server URL makes nats-py verify the server against the system
+        # CA bundle — which trusts the wildcard's Let's Encrypt issuer, so no custom
+        # CA is needed. `token` authenticates this node to the broker.
         self._nc = await nats.connect(
             self._servers,
             name=f"castle-{self._local_hostname}",
+            token=self._token,
             max_reconnect_attempts=-1,  # reconnect forever — nodes come and go
         )
         js = self._nc.jetstream()

docs/fleet-mesh-plan.md

@@ -1,8 +1,9 @@
 # Fleet Mesh Plan — OpenBao + NATS
 
-**Status:** in progress on branch `feat/fleet-mesh-nats-openbao`.
+**Status:** Phases 0–4 complete + verified live across **both nodes** (civil +
-Phases 0–2 + Phase 4 (secret-read backend) complete + single-node verified (live).
+primer), including TLS hardening (NATS + OpenBao). Merged to main.
-Phase 3 (cross-node routing/breaker) and Phase 4 hardening pending the 2nd node.
+Remaining nice-to-have: the live curl+kill breaker demo (needs a peer-unique
+consumed service) — every underlying piece is verified.
 
 ## Context
 
@@ -245,12 +246,23 @@ second node is proven on NATS.
    systemd spec (general, `-`-prefixed so a hook failure never fails the unit);
    `castle-openbao` runs `/data/castle/castle-openbao/unseal.sh` (polls up, unseals
    with `OPENBAO_UNSEAL_KEY`). Verified: manual seal → restart → auto-unsealed.
-3. **TLS hardening — REMAINING (deliberately deferred).** NATS + OpenBao are
+3. **TLS hardening — DONE + verified across both nodes (2026-07-07).**
-   plaintext on the trusted LAN. This is the gate before cross-*network* or moving
+   Reuses the existing ACME **wildcard cert** via `expose.tcp.tls` (the postgres
-   real secrets — neither of which is here yet (the vault holds no production
+   mechanism) — services present a publicly-trusted cert for
-   secrets). Doing NATS TLS/auth touches the *live* civil↔primer mesh, so it's a
+   `<name>.civil.payne.io`, so clients verify against the **system CA with no
-   deliberate, staged change (regenerate certs, update both nodes' clients), not a
+   custom CA to distribute**.
-   tail-end one. Next dedicated step.
+   - **NATS:** `tls{}` (wildcard cert in `/tls`) + `authorization{token}`
+     (`$NATS_TOKEN` from the `NATS_TOKEN` secret). Clients connect to
+     `tls://castle-nats.civil.payne.io:4222` with the token; `CastleNATSClient`
+     gained token support (a `tls://` URL → nats-py verifies via system CA).
+     **civil + primer both cut over**; plaintext now rejected (verified:
+     `certificate is valid for *.civil.payne.io, not localhost`).
+   - **OpenBao:** listener `tls_cert_file`/`tls_key_file` from `/tls`; reachable at
+     `https://castle-openbao.civil.payne.io:8200`. Auto-unseal + secret backend
+     both use the HTTPS URL. Verified: HTTPS serves, **auto-unsealed over HTTPS**,
+     plaintext rejected, backend reads a secret over HTTPS.
+   - Coordinated cutover across the whole fleet (both nodes) with a brief,
+     expected mesh blip; mesh isn't load-bearing so no service impact.
 
 ### Phase 3 — cross-node routing + breaker: logic DONE + verified against a real peer (2026-07-07)