Handle missing streamed token metrics

src/aituner/http_client.py

@@ -310,7 +310,7 @@ def stream_chat_completion(
     return StreamMetrics(
         ttft_ms=ttft_ms,
         tpot_ms=tpot_ms,
-        completion_tokens=used_tokens if used_tokens > 0 else None,
+        completion_tokens=used_tokens if used_tokens is not None and used_tokens > 0 else None,
         completion_tokens_source=completion_tokens_source,
         streamed_chunk_count=chunk_token_count,
     )

src/aituner/worker.py

@@ -8,6 +8,7 @@ import statistics
 import subprocess
 import threading
 import time
+import traceback
 from concurrent.futures import FIRST_COMPLETED, Future, ThreadPoolExecutor, wait
 from dataclasses import dataclass
 from pathlib import Path
@@ -596,6 +597,7 @@ def run_trial(trial_spec_path: Path) -> dict[str, Any]:
                 "best_request_count": None,
                 "failure_stage": failure_stage,
                 "failure_reason": str(exc),
+                "failure_traceback": traceback.format_exc(),
                 "probes": probe_history,
             }
             StudyStore.write_json(Path(trial.result_path), result)

tests/test_core_flow.py

@@ -11,7 +11,13 @@ from unittest import mock
 from aituner.cli import main as cli_main
 from aituner.compare import _aggregate_summary, load_compare_spec, run_compare
 from aituner.engine import build_launch_recipe
-from aituner.http_client import StreamMetrics, _auth_headers, _openai_url, _should_bypass_proxy
+from aituner.http_client import (
+    StreamMetrics,
+    _auth_headers,
+    _openai_url,
+    _should_bypass_proxy,
+    stream_chat_completion,
+)
 from aituner.job import append_job, build_trial_job
 from aituner.harness import (
     build_harness_context,
@@ -3767,6 +3773,29 @@ class CoreFlowTests(unittest.TestCase):
             "http://example.com/v1/chat/completions",
         )
 
+    def test_stream_chat_completion_handles_missing_usage_and_chunks(self) -> None:
+        class FakeResponse:
+            def __enter__(self):
+                return self
+
+            def __exit__(self, exc_type, exc, traceback):
+                return False
+
+            def __iter__(self):
+                return iter([b"data: {\"choices\": []}\n", b"data: [DONE]\n"])
+
+        with mock.patch("aituner.http_client._urlopen", return_value=FakeResponse()):
+            metrics = stream_chat_completion(
+                base_url="http://127.0.0.1:8000",
+                body={"model": "m", "messages": [{"role": "user", "content": "x"}]},
+                timeout_s=1.0,
+            )
+
+        self.assertIsNone(metrics.ttft_ms)
+        self.assertIsNone(metrics.tpot_ms)
+        self.assertIsNone(metrics.completion_tokens)
+        self.assertEqual(metrics.completion_tokens_source, "none")
+
     def test_loopback_urls_bypass_proxy(self) -> None:
         self.assertTrue(_should_bypass_proxy("http://127.0.0.1:8000/v1/models"))
         self.assertTrue(_should_bypass_proxy("http://localhost:8000/health"))