Add datadog health check support + fix bedrock converse cost tracking w/ region name specified (#7958)

* fix(bedrock/converse_handler.py): fix bedrock region name on async calls

* fix(utils.py): fix split model handling

Fixes bedrock cost calculation when region name is given

* feat(_health_endpoints.py): support health checking datadog integration

Closes https://github.com/BerriAI/litellm/issues/7921

litellm/litellm_core_utils/litellm_logging.py

@@ -3341,3 +3341,85 @@ def _get_traceback_str_for_error(error_str: str) -> str:
     function wrapped with lru_cache to limit the number of times `traceback.format_exc()` is called
     """
     return traceback.format_exc()
+
+
+from decimal import Decimal
+
+# used for unit testing
+from typing import Any, Dict, List, Optional, Union
+
+
+def create_dummy_standard_logging_payload() -> StandardLoggingPayload:
+    # First create the nested objects with proper typing
+    model_info = StandardLoggingModelInformation(
+        model_map_key="gpt-3.5-turbo", model_map_value=None
+    )
+
+    metadata = StandardLoggingMetadata(  # type: ignore
+        user_api_key_hash=str("test_hash"),
+        user_api_key_alias=str("test_alias"),
+        user_api_key_team_id=str("test_team"),
+        user_api_key_user_id=str("test_user"),
+        user_api_key_team_alias=str("test_team_alias"),
+        user_api_key_org_id=None,
+        spend_logs_metadata=None,
+        requester_ip_address=str("127.0.0.1"),
+        requester_metadata=None,
+        user_api_key_end_user_id=str("test_end_user"),
+    )
+
+    hidden_params = StandardLoggingHiddenParams(
+        model_id=None,
+        cache_key=None,
+        api_base=None,
+        response_cost=None,
+        additional_headers=None,
+        litellm_overhead_time_ms=None,
+    )
+
+    # Convert numeric values to appropriate types
+    response_cost = Decimal("0.1")
+    start_time = Decimal("1234567890.0")
+    end_time = Decimal("1234567891.0")
+    completion_start_time = Decimal("1234567890.5")
+    saved_cache_cost = Decimal("0.0")
+
+    # Create messages and response with proper typing
+    messages: List[Dict[str, str]] = [{"role": "user", "content": "Hello, world!"}]
+    response: Dict[str, List[Dict[str, Dict[str, str]]]] = {
+        "choices": [{"message": {"content": "Hi there!"}}]
+    }
+
+    # Main payload initialization
+    return StandardLoggingPayload(  # type: ignore
+        id=str("test_id"),
+        call_type=str("completion"),
+        stream=bool(False),
+        response_cost=response_cost,
+        response_cost_failure_debug_info=None,
+        status=str("success"),
+        total_tokens=int(30),
+        prompt_tokens=int(20),
+        completion_tokens=int(10),
+        startTime=start_time,
+        endTime=end_time,
+        completionStartTime=completion_start_time,
+        model_map_information=model_info,
+        model=str("gpt-3.5-turbo"),
+        model_id=str("model-123"),
+        model_group=str("openai-gpt"),
+        custom_llm_provider=str("openai"),
+        api_base=str("https://api.openai.com"),
+        metadata=metadata,
+        cache_hit=bool(False),
+        cache_key=None,
+        saved_cache_cost=saved_cache_cost,
+        request_tags=[],
+        end_user=None,
+        requester_ip_address=str("127.0.0.1"),
+        messages=messages,
+        response=response,
+        error_str=None,
+        model_parameters={"stream": True},
+        hidden_params=hidden_params,
+    )

litellm/litellm_core_utils/logging_utils.py

@@ -110,7 +110,7 @@ def _set_duration_in_model_call_details(
         if logging_obj and hasattr(logging_obj, "model_call_details"):
             logging_obj.model_call_details["llm_api_duration_ms"] = duration_ms
         else:
-            verbose_logger.warning(
+            verbose_logger.debug(
                 "`logging_obj` not found - unable to track `llm_api_duration_ms"
             )
     except Exception as e: