Verify exact production traces and record rollout

apps/ai-service/tests/test_api.py

@@ -1,5 +1,6 @@
 from fastapi.testclient import TestClient
 
+from adapters.prometheus import PrometheusMetrics
 from config import Settings
 from main import create_app
 from rag.agent import AgentReply
@@ -177,3 +178,63 @@ def test_a_trace_write_failure_does_not_turn_a_good_answer_into_a_500():
     assert body["answer"] == "Liều 500 mg [1]."
     assert body["trace_id"]  # a locally-generated fallback id, still present
     assert metrics.total(TRACE_WRITE_FAILED) == 1
+
+
+class _AnyRouting:
+    def retrieve(self, query, subject_scope, intent):
+        return RetrievalResult(EvidenceDecision.ABSTAIN, "drug_not_resolved")
+
+
+def test_correlation_id_round_trips_through_headers_body_and_trace_row():
+    traces = MemoryTraceWriter()
+    app = create_app(
+        settings=Settings(embedding_provider="disabled"),
+        answer_service=GroundedAnswerService(_AnyRouting()),
+        trace_writer=traces,
+    )
+
+    response = TestClient(app).post(
+        "/v1/rag/query",
+        headers={"X-Correlation-ID": "req-test-1"},
+        json={
+            "query": "Paracetamol dose?",
+            "subject_scope": "human",
+            "intent": "fact_lookup",
+        },
+    )
+
+    assert response.status_code == 200
+    assert response.headers["x-correlation-id"] == "req-test-1"
+    assert response.json()["correlation_id"] == "req-test-1"
+    assert traces.rows[0]["correlation_id"] == "req-test-1"
+    assert "otel_trace_id" in traces.rows[0]
+
+
+def test_metrics_endpoint_exposes_request_decision_and_stage_histograms():
+    metrics = PrometheusMetrics()
+    app = create_app(
+        settings=Settings(embedding_provider="disabled"),
+        answer_service=GroundedAnswerService(_AnyRouting()),
+        trace_writer=MemoryTraceWriter(),
+        metrics=metrics,
+    )
+    client = TestClient(app)
+
+    response = client.post(
+        "/v1/rag/query",
+        json={
+            "query": "Paracetamol dose?",
+            "subject_scope": "human",
+            "intent": "fact_lookup",
+        },
+    )
+    assert response.status_code == 200
+
+    scrape = client.get("/metrics")
+    assert scrape.status_code == 200
+    body = scrape.text
+    assert 'duocthu_requests_total{method="POST",route="/v1/rag/query",status="2xx"}' in body
+    assert 'duocthu_decision_total{decision="abstain",reason="drug_not_resolved"}' in body
+    assert 'duocthu_stage_duration_seconds_count{outcome="ok",stage="receive"}' in body
+    assert 'duocthu_stage_duration_seconds_count{outcome="ok",stage="persistence"}' in body
+    assert 'duocthu_stage_duration_seconds_count{outcome="ok",stage="response"}' in body