Merge branch 'main' into weslayer/boto-span-replace

Author: xrmx

CHANGELOG.md

@@ -13,8 +13,14 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- `opentelemetry-instrumentation-system-metrics`: fix loading on Google Cloud Run
+  ([#3533](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/3533))
 - `opentelemetry-instrumentation-fastapi`: fix wrapping of middlewares
   ([#3012](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/3012))
+- `opentelemetry-instrumentation-urllib3`: proper bucket boundaries in stable semconv http duration metrics
+  ([#3518](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/3518))
+- `opentelemetry-instrumentation-urllib`: proper bucket boundaries in stable semconv http duration metrics
+  ([#3519](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/3519))
 - `opentelemetry-instrumentation-falcon`: proper bucket boundaries in stable semconv http duration
   ([#3525](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/3525))
 - `opentelemetry-instrumentation-wsgi`: add explicit http duration buckets for stable semconv
@@ -27,6 +33,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   ([#3524](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/3524))
 - `opentelemetry-instrumentation-grpc`: support non-list interceptors
   ([#3520](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/3520))
+- `opentelemetry-instrumentation-botocore` Ensure spans end on early stream closure for Bedrock Streaming APIs
+  ([#3481](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/3481))
 
 ### Breaking changes
 

instrumentation/opentelemetry-instrumentation-botocore/src/opentelemetry/instrumentation/botocore/extensions/bedrock.py

@@ -499,18 +499,20 @@ def _converse_on_success(
                     [stop_reason],
                 )
 
-        event_logger = instrumentor_context.event_logger
-        choice = _Choice.from_converse(result, capture_content)
-        # this path is used by streaming apis, in that case we are already out of the span
-        # context so need to add the span context manually
-        span_ctx = span.get_span_context()
-        event_logger.emit(
-            choice.to_choice_event(
-                trace_id=span_ctx.trace_id,
-                span_id=span_ctx.span_id,
-                trace_flags=span_ctx.trace_flags,
+        # In case of an early stream closure, the result may not contain outputs
+        if self._stream_has_output_content(result):
+            event_logger = instrumentor_context.event_logger
+            choice = _Choice.from_converse(result, capture_content)
+            # this path is used by streaming apis, in that case we are already out of the span
+            # context so need to add the span context manually
+            span_ctx = span.get_span_context()
+            event_logger.emit(
+                choice.to_choice_event(
+                    trace_id=span_ctx.trace_id,
+                    span_id=span_ctx.span_id,
+                    trace_flags=span_ctx.trace_flags,
+                )
             )
-        )
 
         metrics = instrumentor_context.metrics
         metrics_attributes = self._extract_metrics_attributes()
@@ -602,11 +604,14 @@ def _on_stream_error_callback(
         span: Span,
         exception,
         instrumentor_context: _BotocoreInstrumentorContext,
+        span_ended: bool,
     ):
         span.set_status(Status(StatusCode.ERROR, str(exception)))
         if span.is_recording():
             span.set_attribute(ERROR_TYPE, type(exception).__qualname__)
-        span.end()
+
+        if not span_ended:
+            span.end()
 
         metrics = instrumentor_context.metrics
         metrics_attributes = {
@@ -638,15 +643,17 @@ def on_success(
                 result["stream"], EventStream
             ):
 
-                def stream_done_callback(response):
+                def stream_done_callback(response, span_ended):
                     self._converse_on_success(
                         span, response, instrumentor_context, capture_content
                     )
-                    span.end()
 
-                def stream_error_callback(exception):
+                    if not span_ended:
+                        span.end()
+
+                def stream_error_callback(exception, span_ended):
                     self._on_stream_error_callback(
-                        span, exception, instrumentor_context
+                        span, exception, instrumentor_context, span_ended
                     )
 
                 result["stream"] = ConverseStreamWrapper(
@@ -677,16 +684,17 @@ def stream_error_callback(exception):
         elif self._call_context.operation == "InvokeModelWithResponseStream":
             if "body" in result and isinstance(result["body"], EventStream):
 
-                def invoke_model_stream_done_callback(response):
+                def invoke_model_stream_done_callback(response, span_ended):
                     # the callback gets data formatted as the simpler converse API
                     self._converse_on_success(
                         span, response, instrumentor_context, capture_content
                     )
-                    span.end()
+                    if not span_ended:
+                        span.end()
 
-                def invoke_model_stream_error_callback(exception):
+                def invoke_model_stream_error_callback(exception, span_ended):
                     self._on_stream_error_callback(
-                        span, exception, instrumentor_context
+                        span, exception, instrumentor_context, span_ended
                     )
 
                 result["body"] = InvokeModelWithResponseStreamWrapper(
@@ -781,9 +789,11 @@ def _handle_amazon_nova_response(
                 GEN_AI_RESPONSE_FINISH_REASONS, [response_body["stopReason"]]
             )
 
-        event_logger = instrumentor_context.event_logger
-        choice = _Choice.from_converse(response_body, capture_content)
-        event_logger.emit(choice.to_choice_event())
+        # In case of an early stream closure, the result may not contain outputs
+        if self._stream_has_output_content(response_body):
+            event_logger = instrumentor_context.event_logger
+            choice = _Choice.from_converse(response_body, capture_content)
+            event_logger.emit(choice.to_choice_event())
 
         metrics = instrumentor_context.metrics
         metrics_attributes = self._extract_metrics_attributes()
@@ -1004,3 +1014,8 @@ def on_error(
                 duration,
                 attributes=metrics_attributes,
             )
+
+    def _stream_has_output_content(self, response_body: dict[str, Any]):
+        return (
+            "output" in response_body and "message" in response_body["output"]
+        )

instrumentation/opentelemetry-instrumentation-botocore/src/opentelemetry/instrumentation/botocore/extensions/bedrock_utils.py

@@ -65,14 +65,15 @@ def __init__(
         self._message = None
         self._content_block = {}
         self._record_message = False
+        self._ended = False
 
     def __iter__(self):
         try:
             for event in self.__wrapped__:
                 self._process_event(event)
                 yield event
         except EventStreamError as exc:
-            self._stream_error_callback(exc)
+            self._handle_stream_error(exc)
             raise
 
     def _process_event(self, event):
@@ -133,11 +134,23 @@ def _process_event(self, event):
 
                 if output_tokens := usage.get("outputTokens"):
                     self._response["usage"]["outputTokens"] = output_tokens
-
-            self._stream_done_callback(self._response)
+            self._complete_stream(self._response)
 
             return
 
+    def close(self):
+        self.__wrapped__.close()
+        # Treat the stream as done to ensure the span end.
+        self._complete_stream(self._response)
+
+    def _complete_stream(self, response):
+        self._stream_done_callback(response, self._ended)
+        self._ended = True
+
+    def _handle_stream_error(self, exc):
+        self._stream_error_callback(exc, self._ended)
+        self._ended = True
+
 
 # pylint: disable=abstract-method
 class InvokeModelWithResponseStreamWrapper(ObjectProxy):
@@ -163,14 +176,28 @@ def __init__(
         self._content_block = {}
         self._tool_json_input_buf = ""
         self._record_message = False
+        self._ended = False
+
+    def close(self):
+        self.__wrapped__.close()
+        # Treat the stream as done to ensure the span end.
+        self._stream_done_callback(self._response, self._ended)
+
+    def _complete_stream(self, response):
+        self._stream_done_callback(response, self._ended)
+        self._ended = True
+
+    def _handle_stream_error(self, exc):
+        self._stream_error_callback(exc, self._ended)
+        self._ended = True
 
     def __iter__(self):
         try:
             for event in self.__wrapped__:
                 self._process_event(event)
                 yield event
         except EventStreamError as exc:
-            self._stream_error_callback(exc)
+            self._handle_stream_error(exc)
             raise
 
     def _process_event(self, event):
@@ -213,7 +240,7 @@ def _process_amazon_titan_chunk(self, chunk):
             self._response["output"] = {
                 "message": {"content": [{"text": chunk["outputText"]}]}
             }
-            self._stream_done_callback(self._response)
+            self._complete_stream(self._response)
 
     def _process_amazon_nova_chunk(self, chunk):
         # pylint: disable=too-many-branches
@@ -283,7 +310,7 @@ def _process_amazon_nova_chunk(self, chunk):
                 if output_tokens := usage.get("outputTokens"):
                     self._response["usage"]["outputTokens"] = output_tokens
 
-            self._stream_done_callback(self._response)
+            self._complete_stream(self._response)
             return
 
     def _process_anthropic_claude_chunk(self, chunk):
@@ -355,7 +382,7 @@ def _process_anthropic_claude_chunk(self, chunk):
                 self._record_message = False
                 self._message = None
 
-            self._stream_done_callback(self._response)
+            self._complete_stream(self._response)
             return
 
 

instrumentation/opentelemetry-instrumentation-botocore/tests/cassettes/test_converse_stream_close_before_consumption.yaml

@@ -0,0 +1,81 @@
+interactions:
+- request:
+    body: |-
+      {
+        "messages": [
+          {
+            "role": "user",
+            "content": [
+              {
+                "text": "Say this is a test"
+              }
+            ]
+          }
+        ],
+        "inferenceConfig": {
+          "maxTokens": 10,
+          "temperature": 0.8,
+          "topP": 1,
+          "stopSequences": [
+            "|"
+          ]
+        }
+      }
+    headers:
+      Content-Length:
+      - '170'
+      Content-Type:
+      - application/json
+      User-Agent:
+      - Boto3/1.35.56 md/Botocore#1.35.56 ua/2.0 os/macos#24.4.0 md/arch#arm64 lang/python#3.12.0
+        md/pyimpl#CPython cfg/retry-mode#legacy Botocore/1.35.56
+      X-Amz-Date:
+      - 20250509T104610Z
+      X-Amz-Security-Token:
+      - test_aws_security_token
+      X-Amzn-Trace-Id:
+      - Root=1-98fd1391-6f6d1be3cc209d9ba2fb3034;Parent=6dc095915efed44e;Sampled=1
+      amz-sdk-invocation-id:
+      - 3a37ec88-d989-4397-bfbe-d463b66a3a08
+      amz-sdk-request:
+      - attempt=1
+      authorization:
+      - Bearer test_aws_authorization
+    method: POST
+    uri: https://bedrock-runtime.us-east-1.amazonaws.com/model/amazon.titan-text-lite-v1/converse-stream
+  response:
+    body:
+      string: !!binary |
+        AAAApAAAAFJl4NbnCzpldmVudC10eXBlBwAMbWVzc2FnZVN0YXJ0DTpjb250ZW50LXR5cGUHABBh
+        cHBsaWNhdGlvbi9qc29uDTptZXNzYWdlLXR5cGUHAAVldmVudHsicCI6ImFiY2RlZmdoaWprbG1u
+        b3BxcnN0dXZ3eHl6QUJDREVGR0hJSktMTSIsInJvbGUiOiJhc3Npc3RhbnQifSKBqDQAAADZAAAA
+        VxTIBhYLOmV2ZW50LXR5cGUHABFjb250ZW50QmxvY2tEZWx0YQ06Y29udGVudC10eXBlBwAQYXBw
+        bGljYXRpb24vanNvbg06bWVzc2FnZS10eXBlBwAFZXZlbnR7ImNvbnRlbnRCbG9ja0luZGV4Ijow
+        LCJkZWx0YSI6eyJ0ZXh0IjoiSGksIGRpZCB5b3UgaGF2ZSBhIHF1ZXN0aW9uIn0sInAiOiJhYmNk
+        ZWZnaGlqa2xtbm9wcXJzdHV2d3h5ekFCQ0RFRkdISUpLIn22h6U3AAAAqgAAAFbdvayfCzpldmVu
+        dC10eXBlBwAQY29udGVudEJsb2NrU3RvcA06Y29udGVudC10eXBlBwAQYXBwbGljYXRpb24vanNv
+        bg06bWVzc2FnZS10eXBlBwAFZXZlbnR7ImNvbnRlbnRCbG9ja0luZGV4IjowLCJwIjoiYWJjZGVm
+        Z2hpamtsbW5vcHFyc3R1dnd4eXpBQkNERUZHSElKS0wifQSjvQkAAAChAAAAUTQJCC0LOmV2ZW50
+        LXR5cGUHAAttZXNzYWdlU3RvcA06Y29udGVudC10eXBlBwAQYXBwbGljYXRpb24vanNvbg06bWVz
+        c2FnZS10eXBlBwAFZXZlbnR7InAiOiJhYmNkZWZnaGlqa2xtbm9wcXJzdHV2d3h5ekFCQ0QiLCJz
+        dG9wUmVhc29uIjoibWF4X3Rva2VucyJ9jxPapgAAAMsAAABOaoNqNAs6ZXZlbnQtdHlwZQcACG1l
+        dGFkYXRhDTpjb250ZW50LXR5cGUHABBhcHBsaWNhdGlvbi9qc29uDTptZXNzYWdlLXR5cGUHAAVl
+        dmVudHsibWV0cmljcyI6eyJsYXRlbmN5TXMiOjYyNX0sInAiOiJhYmNkZWZnaGlqa2wiLCJ1c2Fn
+        ZSI6eyJpbnB1dFRva2VucyI6OCwib3V0cHV0VG9rZW5zIjoxMCwidG90YWxUb2tlbnMiOjE4fX0B
+        /Sts
+    headers:
+      Connection:
+      - keep-alive
+      Content-Type:
+      - application/vnd.amazon.eventstream
+      Date:
+      - Fri, 09 May 2025 10:46:10 GMT
+      Set-Cookie: test_set_cookie
+      Transfer-Encoding:
+      - chunked
+      x-amzn-RequestId:
+      - 5aaaa521-d2c6-4980-ba0a-7bba19633a40
+    status:
+      code: 200
+      message: OK
+version: 1