feat - otel log /team exceptions

litellm/integrations/opentelemetry.py

@@ -613,3 +613,46 @@ class OpenTelemetry(CustomLogger):
                     management_endpoint_span.set_attribute(f"response.{key}", value)
             management_endpoint_span.set_status(Status(StatusCode.OK))
             management_endpoint_span.end(end_time=_end_time_ns)
+
+    async def async_management_endpoint_failure_hook(
+        self,
+        logging_payload: ManagementEndpointLoggingPayload,
+        parent_otel_span: Optional[Span] = None,
+    ):
+        from opentelemetry import trace
+        from datetime import datetime
+        from opentelemetry.trace import Status, StatusCode
+
+        _start_time_ns = logging_payload.start_time
+        _end_time_ns = logging_payload.end_time
+
+        start_time = logging_payload.start_time
+        end_time = logging_payload.end_time
+
+        if isinstance(start_time, float):
+            _start_time_ns = int(int(start_time) * 1e9)
+        else:
+            _start_time_ns = self._to_ns(start_time)
+
+        if isinstance(end_time, float):
+            _end_time_ns = int(int(end_time) * 1e9)
+        else:
+            _end_time_ns = self._to_ns(end_time)
+
+        if parent_otel_span is not None:
+            _span_name = logging_payload.route
+            management_endpoint_span = self.tracer.start_span(
+                name=_span_name,
+                context=trace.set_span_in_context(parent_otel_span),
+                start_time=_start_time_ns,
+            )
+
+            _request_data = logging_payload.request_data
+            if _request_data is not None:
+                for key, value in _request_data.items():
+                    management_endpoint_span.set_attribute(f"request.{key}", value)
+
+            _exception = logging_payload.exception
+            management_endpoint_span.set_attribute(f"exception", str(_exception))
+            management_endpoint_span.set_status(Status(StatusCode.ERROR))
+            management_endpoint_span.end(end_time=_end_time_ns)

litellm/proxy/common_utils/management_endpoint_utils.py

@@ -1,7 +1,7 @@
 from datetime import datetime
 from functools import wraps
 from litellm.proxy._types import UserAPIKeyAuth, ManagementEndpointLoggingPayload
-from http_parsing_utils import _read_request_body
+from common_utils.http_parsing_utils import _read_request_body
 from fastapi import Request
 
 
@@ -20,6 +20,7 @@ def management_endpoint_wrapper(func):
         try:
             result = await func(*args, **kwargs)
             end_time = datetime.now()
+
             if kwargs is None:
                 kwargs = {}
             user_api_key_dict: UserAPIKeyAuth = kwargs.get("user_api_key_dict")
@@ -52,6 +53,34 @@ def management_endpoint_wrapper(func):
             return result
         except Exception as e:
             end_time = datetime.now()
+
+            if kwargs is None:
+                kwargs = {}
+            user_api_key_dict: UserAPIKeyAuth = kwargs.get("user_api_key_dict")
+            parent_otel_span = user_api_key_dict.parent_otel_span
+            if parent_otel_span is not None:
+                from litellm.proxy.proxy_server import open_telemetry_logger
+
+                if open_telemetry_logger is not None:
+                    _http_request: Request = kwargs.get("http_request")
+                    _route = _http_request.url.path
+                    _request_body: dict = await _read_request_body(
+                        request=_http_request
+                    )
+                    logging_payload = ManagementEndpointLoggingPayload(
+                        route=_route,
+                        request_data=_request_body,
+                        response=None,
+                        start_time=start_time,
+                        end_time=end_time,
+                        exception=e,
+                    )
+
+                    await open_telemetry_logger.async_management_endpoint_failure_hook(
+                        logging_payload=logging_payload,
+                        parent_otel_span=parent_otel_span,
+                    )
+
             raise e
 
     return wrapper

litellm/proxy/proxy_server.py

@@ -114,7 +114,7 @@ from litellm.proxy.utils import (
     _to_ns,
     get_error_message_str,
 )
-from litellm.proxy.utils.http_parsing_utils import _read_request_body
+from litellm.proxy.common_utils.http_parsing_utils import _read_request_body
 
 from litellm import (
     CreateBatchRequest,
@@ -164,7 +164,9 @@ from litellm.proxy.auth.auth_checks import (
     get_actual_routes,
     log_to_opentelemetry,
 )
-from litellm.proxy.utils.management_endpoint_utils import management_endpoint_wrapper
+from litellm.proxy.common_utils.management_endpoint_utils import (
+    management_endpoint_wrapper,
+)
 from litellm.llms.custom_httpx.httpx_handler import HTTPHandler
 from litellm.exceptions import RejectedRequestError
 from litellm.integrations.slack_alerting import SlackAlertingArgs, SlackAlerting