Merge branch 'main' into openai-v2-request-seed

Author: xrmx

CHANGELOG.md

@@ -11,8 +11,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## Unreleased
 
-### Fixed
+### Fixed 
 
+- `opentelemetry-instrumentation-django`: Fix exemplars generation for `http.server.(request.)duration`
+  ([#3945](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/3945))
 - `opentelemetry-util-http`, `opentelemetry-instrumentation-requests`, `opentelemetry-instrumentation-wsgi`, `opentelemetry-instrumentation-asgi`: normalize byte-valued user-agent headers before detecting synthetic sources so attributes are recorded reliably.
   ([#4001](https://github.com/open-telemetry/opentelemetry-python-contrib/issues/4001))
 

instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/cassettes/test_chat_completion_with_context_manager_streaming.yaml

@@ -0,0 +1,141 @@
+interactions:
+- request:
+    body: |-
+      {
+        "messages": [
+          {
+            "role": "user",
+            "content": "Say this is a test"
+          }
+        ],
+        "model": "gpt-4o-mini",
+        "stream": true,
+        "stream_options": {
+          "include_usage": true
+        }
+      }
+    headers:
+      Accept:
+      - application/json
+      Accept-Encoding:
+      - gzip, deflate
+      Connection:
+      - keep-alive
+      Content-Length:
+      - '148'
+      Content-Type:
+      - application/json
+      Host:
+      - api.openai.com
+      User-Agent:
+      - OpenAI/Python 1.109.1
+      X-Stainless-Arch:
+      - x64
+      X-Stainless-Async:
+      - 'false'
+      X-Stainless-Lang:
+      - python
+      X-Stainless-OS:
+      - Linux
+      X-Stainless-Package-Version:
+      - 1.109.1
+      X-Stainless-Runtime:
+      - CPython
+      X-Stainless-Runtime-Version:
+      - 3.12.12
+      authorization:
+      - Bearer test_openai_api_key
+      x-stainless-read-timeout:
+      - '600'
+      x-stainless-retry-count:
+      - '0'
+    method: POST
+    uri: https://api.openai.com/v1/chat/completions
+  response:
+    body:
+      string: |+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"role":"assistant","content":"","refusal":null},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"yKkTGHRZw"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":"This"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"bFd3AyW"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":" is"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"LqISoMXL"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":" a"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"mUrlylOKo"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":" test"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"nGqQio"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":"."},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"gUhadpfoFR"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":" How"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"atLjmXM"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":" can"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"Ac67p8z"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":" I"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"tl4DibKXq"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":" assist"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"p5ei"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":" you"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"97qAT5t"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":" further"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"ZK4"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{"content":"?"},"logprobs":null,"finish_reason":null}],"usage":null,"obfuscation":"5iGjeBvFWd"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[{"index":0,"delta":{},"logprobs":null,"finish_reason":"stop"}],"usage":null,"obfuscation":"SdS98"}
+
+        data: {"id":"chatcmpl-CnMGGNmSdY4yYxO9NmV7cSVxXZ5ci","object":"chat.completion.chunk","created":1765879680,"model":"gpt-4o-mini-2024-07-18","service_tier":"default","system_fingerprint":"fp_644f11dd4d","choices":[],"usage":{"prompt_tokens":12,"completion_tokens":12,"total_tokens":24,"prompt_tokens_details":{"cached_tokens":0,"audio_tokens":0},"completion_tokens_details":{"reasoning_tokens":0,"audio_tokens":0,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"obfuscation":"parVcHlnsA"}
+
+        data: [DONE]
+
+    headers:
+      CF-RAY:
+      - 9aed607a4c274be7-MXP
+      Connection:
+      - keep-alive
+      Content-Type:
+      - text/event-stream; charset=utf-8
+      Date:
+      - Tue, 16 Dec 2025 10:08:00 GMT
+      Server:
+      - cloudflare
+      Set-Cookie: test_set_cookie
+      Strict-Transport-Security:
+      - max-age=31536000; includeSubDomains; preload
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - nosniff
+      access-control-expose-headers:
+      - X-Request-ID
+      alt-svc:
+      - h3=":443"; ma=86400
+      cf-cache-status:
+      - DYNAMIC
+      openai-organization: test_openai_org_id
+      openai-processing-ms:
+      - '274'
+      openai-project:
+      - proj_Pf1eM5R55Z35wBy4rt8PxAGq
+      openai-version:
+      - '2020-10-01'
+      x-envoy-upstream-service-time:
+      - '618'
+      x-openai-proxy-wasm:
+      - v0.1
+      x-ratelimit-limit-requests:
+      - '10000'
+      x-ratelimit-limit-tokens:
+      - '10000000'
+      x-ratelimit-remaining-requests:
+      - '9999'
+      x-ratelimit-remaining-tokens:
+      - '9999993'
+      x-ratelimit-reset-requests:
+      - 6ms
+      x-ratelimit-reset-tokens:
+      - 0s
+      x-request-id:
+      - req_50a8341c02184cbd8b364182eb40ed38
+    status:
+      code: 200
+      message: OK
+version: 1

instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/test_chat_completions.py

@@ -955,6 +955,58 @@ def test_chat_completion_with_content_span_unsampled(
     assert logs[0].log_record.trace_flags == logs[1].log_record.trace_flags
 
 
+@pytest.mark.vcr()
+def test_chat_completion_with_context_manager_streaming(
+    span_exporter, log_exporter, openai_client, instrument_with_content
+):
+    llm_model_value = "gpt-4o-mini"
+    messages_value = [{"role": "user", "content": "Say this is a test"}]
+    with openai_client.chat.completions.create(
+        messages=messages_value,
+        model=llm_model_value,
+        stream=True,
+        stream_options={"include_usage": True},
+    ) as response:
+        message_content = ""
+        for chunk in response:
+            if chunk.choices:
+                message_content += chunk.choices[0].delta.content or ""
+            # get the last chunk
+            if getattr(chunk, "usage", None):
+                response_stream_usage = chunk.usage
+                response_stream_model = chunk.model
+                response_stream_id = chunk.id
+
+    spans = span_exporter.get_finished_spans()
+    assert_all_attributes(
+        spans[0],
+        llm_model_value,
+        response_stream_id,
+        response_stream_model,
+        response_stream_usage.prompt_tokens,
+        response_stream_usage.completion_tokens,
+        response_service_tier="default",
+    )
+
+    logs = log_exporter.get_finished_logs()
+    assert len(logs) == 2
+
+    user_message = {"content": messages_value[0]["content"]}
+    assert_message_in_logs(
+        logs[0], "gen_ai.user.message", user_message, spans[0]
+    )
+
+    choice_event = {
+        "index": 0,
+        "finish_reason": "stop",
+        "message": {
+            "role": "assistant",
+            "content": message_content,
+        },
+    }
+    assert_message_in_logs(logs[1], "gen_ai.choice", choice_event, spans[0])
+
+
 def chat_completion_multiple_tools_streaming(
     span_exporter, log_exporter, openai_client, expect_content
 ):

instrumentation/opentelemetry-instrumentation-django/src/opentelemetry/instrumentation/django/middleware/otel_middleware.py

@@ -417,15 +417,6 @@ def process_response(self, request, response):
                 except Exception:  # pylint: disable=broad-exception-caught
                     _logger.exception("Exception raised by response_hook")
 
-            if exception:
-                activation.__exit__(
-                    type(exception),
-                    exception,
-                    getattr(exception, "__traceback__", None),
-                )
-            else:
-                activation.__exit__(None, None, None)
-
         if request_start_time is not None:
             duration_s = default_timer() - request_start_time
             if self._duration_histogram_old:
@@ -437,16 +428,29 @@ def process_response(self, request, response):
                 if target:
                     duration_attrs_old[SpanAttributes.HTTP_TARGET] = target
                 self._duration_histogram_old.record(
-                    max(round(duration_s * 1000), 0), duration_attrs_old
+                    max(round(duration_s * 1000), 0),
+                    duration_attrs_old,
                 )
             if self._duration_histogram_new:
                 duration_attrs_new = _parse_duration_attrs(
                     duration_attrs, _StabilityMode.HTTP
                 )
                 self._duration_histogram_new.record(
-                    max(duration_s, 0), duration_attrs_new
+                    max(duration_s, 0),
+                    duration_attrs_new,
                 )
         self._active_request_counter.add(-1, active_requests_count_attrs)
+
+        if activation and span:
+            if exception:
+                activation.__exit__(
+                    type(exception),
+                    exception,
+                    getattr(exception, "__traceback__", None),
+                )
+            else:
+                activation.__exit__(None, None, None)
+
         if request.META.get(self._environ_token, None) is not None:
             detach(request.META.get(self._environ_token))
             request.META.pop(self._environ_token)

instrumentation/opentelemetry-instrumentation-django/tests/test_middleware.py

@@ -1104,3 +1104,66 @@ def test_http_custom_response_headers_not_in_span_attributes(self):
         for key, _ in not_expected.items():
             self.assertNotIn(key, span.attributes)
         self.memory_exporter.clear()
+
+
+class TestMiddlewareSpanActivationTiming(WsgiTestBase):
+    """Test span activation timing relative to metrics recording."""
+
+    @classmethod
+    def setUpClass(cls):
+        conf.settings.configure(ROOT_URLCONF=modules[__name__])
+        super().setUpClass()
+
+    def setUp(self):
+        super().setUp()
+        setup_test_environment()
+        _django_instrumentor.instrument()
+
+    def tearDown(self):
+        super().tearDown()
+        teardown_test_environment()
+        _django_instrumentor.uninstrument()
+
+    @classmethod
+    def tearDownClass(cls):
+        super().tearDownClass()
+        conf.settings = conf.LazySettings()
+
+    def test_span_ended_after_metrics_recorded(self):
+        """Span activation exits after metrics recording."""
+        Client().get("/traced/")
+
+        spans = self.memory_exporter.get_finished_spans()
+        self.assertEqual(len(spans), 1)
+
+        # Span properly finished
+        self.assertIsNotNone(spans[0].end_time)
+
+        # Metrics recorded
+        metrics_list = self.memory_metrics_reader.get_metrics_data()
+        histogram_found = any(
+            "duration" in metric.name
+            for rm in metrics_list.resource_metrics
+            for sm in rm.scope_metrics
+            for metric in sm.metrics
+        )
+        self.assertTrue(histogram_found)
+
+    def test_metrics_recorded_with_exception(self):
+        """Metrics recorded even when request raises exception."""
+        with self.assertRaises(ValueError):
+            Client().get("/error/")
+
+        spans = self.memory_exporter.get_finished_spans()
+        self.assertEqual(len(spans), 1)
+        self.assertEqual(spans[0].status.status_code, StatusCode.ERROR)
+
+        # Metrics still recorded
+        metrics_list = self.memory_metrics_reader.get_metrics_data()
+        histogram_found = any(
+            "duration" in metric.name
+            for rm in metrics_list.resource_metrics
+            for sm in rm.scope_metrics
+            for metric in sm.metrics
+        )
+        self.assertTrue(histogram_found)