fix(gateway): grant sandboxclaims patch and map API errors to 503 (#45)

* fix(gateway): grant patch on sandboxclaims for token rotation

The SSO landing path calls K8sClient.patch_token(), which patches the
`dsh-<user>` SandboxClaim's token-hash annotation whenever the browser has
no valid gateway session: first visit, an expired or foreign cookie, or a
gateway restart that drops the in-memory session store. The gateway Role
only granted get/list/create/delete on sandboxclaims, so the patch returned
403 and, because ApiException is not handled, every such GET / answered a
bare 500 "Internal Server Error".

Grant the missing `patch` verb on
extensions.agents.x-k8s.io/sandboxclaims, and update the least-privilege
comment plus the two docs that enumerate the Role verbs.

Verified with `make check` (versions, example layout, manifests) and
`yamllint`; the 403 and the resulting 500 were reproduced from a running
install and cleared by applying the equivalent Role patch (RBAC is evaluated
per request, so no gateway restart is needed).

* fix(gateway): return 503 instead of a bare 500 on Kubernetes API errors

Every gateway route talks to the Kubernetes API, and ApiException used to be
unhandled: Starlette's ServerErrorMiddleware turned it into a bare
500 "Internal Server Error" plus an ERROR traceback. That masked a missing
RBAC verb as a client-visible crash on GET / (the SSO landing path), and it
would equally hide an API outage or throttling.

Register an ApiException handler that logs one credential-free line
(status/reason) and answers 503 + Retry-After with a fixed body, so the
Kubernetes Status payload (which can name cluster objects) is never proxied.
Guard the first get_claim() in the WebSocket route the same way: WebSocket
scopes bypass Starlette's HTTP exception handlers, so it now closes with 1011
instead of escaping as a traceback.

Tests assert the 503 + Retry-After response, cover the SSO landing path that
produced the reported 500 (forbidden claim patch), and assert the WebSocket
close code. The failure-semantics tables document the new behaviour.

---------

Co-authored-by: SREDevOpsAgents <agents@sredevops.org>

examples/dsh-agents-as-a-service/docs/architecture.md

@@ -339,7 +339,7 @@ The gateway runs with a namespace-scoped ServiceAccount (`manifests/50-gateway.y
 
 | API group | Resource | Verbs | Why |
 | --- | --- | --- | --- |
-| `extensions.agents.x-k8s.io` | `sandboxclaims` | `get`, `list`, `create`, `delete` | signup, state, account deletion |
+| `extensions.agents.x-k8s.io` | `sandboxclaims` | `get`, `list`, `create`, `patch`, `delete` | signup, state, account deletion, token rotation |
 | `agents.x-k8s.io` | `sandboxes` | `get`, `list`, `patch` | resolve DNS/pod, wake/suspend |
 | `""` (core) | `pods` | `get`, `list` | resolve the pod behind a sandbox |
 | `""` (core) | `pods/log` | `get` | launch-token handshake |
@@ -404,6 +404,7 @@ Notes and limits:
 | DSH log line changed or rotated away | Proxy returns `503` (missing launch token) | Fix the parser or restart the pod so the token line is re-emitted |
 | Probe on `/` | Pod `CrashLoopBackOff`, `401` in probe logs | Use `/favicon.svg` for readiness/liveness (shipped templates do) |
 | Warm pool exhausted | JSON signup returns `202` (state `Waking`); form signup is slow | Increase `replicas` in the warm pool |
+| Kubernetes API error (API down or throttled, missing RBAC verb) | HTTP `503` + `Retry-After`; WebSocket close `1011` | One credential-free log line with the API `status`/`reason`, plus a fixed body; the Kubernetes `Status` payload is never proxied. Check the ServiceAccount Role and the API server — RBAC is evaluated per request, so a Role fix needs no restart |
 | Gateway restart | Idle sandboxes stop suspending; handshake cache and browser sessions lost | Idle clocks and handshake cache rebuild lazily; sessions do not, so browsers must sign in again after a rollout. Use a shared session store for production |
 | Gateway configured `COOKIE_SECURE=false` without `ALLOW_INSECURE_COOKIES=1` | Gateway pod exits at startup / `CrashLoopBackOff` | Intentional guard against an insecure cookie on an HTTPS deployment; set `ALLOW_INSECURE_COOKIES=1` only for the local plain-HTTP demo, or set `COOKIE_SECURE=true` |
 | Auth rate limit hit | `429` + `Retry-After` on signup/login | Per-client in-memory bucket (`AUTH_RATE_LIMIT`/`AUTH_RATE_WINDOW`); wait for the window or tune it. Behind a proxy, keying uses the rightmost `X-Forwarded-For` entry |

examples/dsh-agents-as-a-service/gateway/README.md

@@ -175,9 +175,11 @@ oauth2-proxy's egress to the gateway, DNS and public HTTPS.
 | Sandbox not provisioned, wake timeout, handshake failure | `503` + `Retry-After: 10`. The handshake error body is a fixed message: it never contains the launch token, the handshake URL or pod/credential details. |
 | User over `MAX_IN_FLIGHT_PER_USER` | `503` + `Retry-After` (HTTP); the existing in-flight requests are untouched |
 | Upstream connection/protocol failure | `502` |
+| Kubernetes API error (API unreachable/throttled, RBAC `403`, `5xx`) | `503` + `Retry-After: 10`. The body is a fixed message; the Kubernetes `Status` payload (which can name cluster objects) is never proxied, and the API `status`/`reason` is logged once |
 | WS auth failure (unknown user, bad token, cross-origin) | accept, then close `4401` |
 | WS per-user concurrency cap | accept, then close `1011` |
 | WS upstream/handshake failure | accept, then close `1011` |
+| WS Kubernetes API error | accept, then close `1011` |
 | WS message from DSH over `WS_MAX_MESSAGE_BYTES` | the upstream/relay closes the connection (websockets' message-too-big path) |
 
 A close code can only be delivered after the WebSocket upgrade is accepted;
@@ -230,7 +232,8 @@ would never be sent). HTTPS deployments must set `COOKIE_SECURE=true`.
 
 The gateway's ServiceAccount needs, in its namespace:
 
-* `sandboxclaims` (`extensions.agents.x-k8s.io`): get, list, create, delete
+* `sandboxclaims` (`extensions.agents.x-k8s.io`): get, list, create, patch, delete
+  (patch rotates the token-hash annotation when a session is minted)
 * `sandboxes` (`agents.x-k8s.io`): get, list, patch
 * `pods`: get, list
 * `pods/log`: get

examples/dsh-agents-as-a-service/gateway/gateway.py

@@ -1510,7 +1510,13 @@ against the JSON API. The browser session is an HttpOnly cookie.</p>
         user = websocket.path_params["user"]
         if not valid_user(user):
             return await self._ws_reject(websocket, 4401)
+        try:
             claim = await self.k8s.get_claim(user)
+        except ApiException:
+            # Starlette has no exception handler for WebSocket scopes, so an
+            # unhandled API error would escape as a traceback and an unclean
+            # close.  1011 keeps the HTTP 503 meaning on the WS channel.
+            return await self._ws_reject(websocket, 1011)
         if claim is None:
             return await self._ws_reject(websocket, 4401)
         token, cookie_user = self.credentials(websocket.headers, websocket.cookies)
@@ -1746,6 +1752,28 @@ async def client_gone(request: Request, exc: ClientDisconnect) -> Response:
     return Response(status_code=CLIENT_CLOSED_REQUEST)
 
 
+async def k8s_api_error(request: Request, exc: ApiException) -> Response:
+    """Answer a Kubernetes API failure with 503 instead of a bare 500.
+
+    The gateway is the only path between the browser and the cluster, so an
+    API outage, a throttle or a missing RBAC verb is a server-side
+    availability problem, not a client error.  A 503 keeps ingress and
+    browser retry semantics honest, and the body is a fixed message: the
+    Kubernetes ``Status`` payload can name cluster objects and is never
+    proxied.  The API ``status``/``reason`` is logged once, credential-free.
+    """
+    print(
+        "gateway: kubernetes API error: "
+        f"status={getattr(exc, 'status', None)} reason={getattr(exc, 'reason', None)}",
+        flush=True,
+    )
+    return JSONResponse(
+        {"error": "cluster API unavailable, retry shortly"},
+        status_code=503,
+        headers={"Retry-After": "10"},
+    )
+
+
 def create_app(
     settings: Settings | None = None,
     k8s=None,
@@ -1778,7 +1806,7 @@ def create_app(
     ]
     app = Starlette(
         routes=routes,
-        exception_handlers={ClientDisconnect: client_gone},
+        exception_handlers={ClientDisconnect: client_gone, ApiException: k8s_api_error},
         lifespan=_lifespan(gateway, http_client if owns_http_client else None),
     )
     app.state.gateway = gateway

examples/dsh-agents-as-a-service/gateway/test_gateway.py

@@ -1176,12 +1176,14 @@ async def test_delete_missing_claim_is_404(env):
 
 
 @pytest.mark.asyncio
-async def test_unexpected_api_error_propagates(env):
+async def test_unexpected_api_error_returns_503(env):
     add_user(env.k8s)
     env.k8s.get_claim_error = ApiException(status=500)
     async with asgi_client(env) as client:
-        with pytest.raises(ApiException):
+        response = await client.get("/users/alice")
-            await client.get("/users/alice")
+    assert response.status_code == 503
+    assert response.headers["retry-after"] == "10"
+    assert response.json() == {"error": "cluster API unavailable, retry shortly"}
 
 
 @pytest.mark.asyncio
@@ -1831,6 +1833,16 @@ def test_ws_unauthenticated_closes_4401(env):
     assert exc.value.code == 4401
 
 
+def test_ws_kubernetes_api_error_closes_1011(env):
+    add_user(env.k8s)
+    env.k8s.get_claim_error = ApiException(status=503)
+    with TestClient(env.app) as client:
+        with client.websocket_connect("/u/alice/api/remote.mux") as ws:
+            with pytest.raises(WebSocketDisconnect) as exc:
+                ws.receive_text()
+    assert exc.value.code == 1011
+
+
 def test_ws_cookie_cross_origin_closes_4401(env):
     add_user(env.k8s)
     with TestClient(env.app) as client:
@@ -2149,6 +2161,19 @@ def test_remote_user_returns_valid_and_rejects_invalid(env):
     assert g.remote_user({}) is None
 
 
+@pytest.mark.asyncio
+async def test_sso_landing_returns_503_when_claim_patch_is_forbidden(env):
+    """A missing RBAC verb on ``patch sandboxclaims`` used to escape as a 500."""
+    enable_sso(env)
+    env.k8s.claims["alice"] = {"metadata": {"name": claim_name("alice")}}
+    env.k8s.patch_error = ApiException(status=403)
+    async with asgi_client(env) as client:
+        response = await client.get("/", headers={"X-Forwarded-User": "alice"})
+    assert response.status_code == 503
+    assert response.headers["retry-after"] == "10"
+    assert response.json() == {"error": "cluster API unavailable, retry shortly"}
+
+
 @pytest.mark.asyncio
 async def test_sso_landing_new_user_shows_provisioning(env):
     enable_sso(env)

examples/dsh-agents-as-a-service/manifests/50-gateway.yaml

@@ -26,7 +26,8 @@ metadata:
   namespace: dsh-demo
 ---
 # Least privilege:
-#   - claims are created/deleted per user (signup/delete);
+#   - claims are created/deleted per user (signup/delete) and patched only to
+#     rotate the token-hash annotation when a session is minted;
 #   - sandboxes are read, and patched only to flip operatingMode (suspend/resume);
 #   - pods are read, and pods/log is read to capture DSH's launch token from
 #     the startup log line (`dsh web: ...?token=...`).
@@ -39,7 +40,7 @@ metadata:
 rules:
   - apiGroups: ["extensions.agents.x-k8s.io"]
     resources: ["sandboxclaims"]
-    verbs: ["get", "list", "create", "delete"]
+    verbs: ["get", "list", "create", "patch", "delete"]
   - apiGroups: ["agents.x-k8s.io"]
     resources: ["sandboxes"]
     verbs: ["get", "list", "patch"]