fix simplify - pass litellm_parent_otel_span

litellm/integrations/opentelemetry.py

@@ -162,7 +162,8 @@ class OpenTelemetry(CustomLogger):
         proxy_server_request = litellm_params.get("proxy_server_request", {}) or {}
         headers = proxy_server_request.get("headers", {}) or {}
         traceparent = headers.get("traceparent", None)
-        parent_otel_span = litellm_params.get("litellm_parent_otel_span", None)
+        _metadata = litellm_params.get("metadata", {})
+        parent_otel_span = _metadata.get("litellm_parent_otel_span", None)
 
         """
         Two way to use parents in opentelemetry

litellm/main.py

@@ -600,7 +600,6 @@ def completion(
     client = kwargs.get("client", None)
     ### Admin Controls ###
     no_log = kwargs.get("no-log", False)
-    litellm_parent_otel_span = kwargs.get("litellm_parent_otel_span", None)
     ######## end of unpacking kwargs ###########
     openai_params = [
         "functions",
@@ -690,7 +689,6 @@ def completion(
         "allowed_model_region",
         "model_config",
         "fastest_response",
-        "litellm_parent_otel_span",
     ]
 
     default_params = openai_params + litellm_params
@@ -875,7 +873,6 @@ def completion(
             input_cost_per_token=input_cost_per_token,
             output_cost_per_second=output_cost_per_second,
             output_cost_per_token=output_cost_per_token,
-            litellm_parent_otel_span=litellm_parent_otel_span,
         )
         logging.update_environment_variables(
             model=model,

litellm/proxy/litellm_pre_call_utils.py

@@ -106,7 +106,7 @@ async def add_litellm_data_to_request(
     data["metadata"]["headers"] = _headers
     data["metadata"]["endpoint"] = str(request.url)
     # Add the OTEL Parent Trace before sending it LiteLLM
-    data["litellm_parent_otel_span"] = user_api_key_dict.parent_otel_span
+    data["metadata"]["litellm_parent_otel_span"] = user_api_key_dict.parent_otel_span
 
     ### END-USER SPECIFIC PARAMS ###
     if user_api_key_dict.allowed_model_region is not None:

litellm/proxy/utils.py

@@ -73,7 +73,8 @@ def print_verbose(print_statement):
 def safe_deep_copy(data):
     if isinstance(data, dict):
         # remove litellm_parent_otel_span since this is not picklable
-        data.pop("litellm_parent_otel_span", None)
+        if "metadata" in data and "litellm_parent_otel_span" in data["metadata"]:
+            data["metadata"].pop("litellm_parent_otel_span")
     new_data = copy.deepcopy(data)
     return new_data
 

litellm/tests/test_proxy_server.py

@@ -152,7 +152,6 @@ def test_chat_completion(mock_acompletion, client_no_auth):
             specific_deployment=True,
             metadata=mock.ANY,
             proxy_server_request=mock.ANY,
-            litellm_parent_otel_span=mock.ANY,
         )
         print(f"response - {response.text}")
         assert response.status_code == 200

litellm/utils.py

@@ -4918,7 +4918,6 @@ def get_litellm_params(
     input_cost_per_token=None,
     output_cost_per_token=None,
     output_cost_per_second=None,
-    litellm_parent_otel_span=None,
 ):
     litellm_params = {
         "acompletion": acompletion,
@@ -4941,7 +4940,6 @@ def get_litellm_params(
         "input_cost_per_second": input_cost_per_second,
         "output_cost_per_token": output_cost_per_token,
         "output_cost_per_second": output_cost_per_second,
-        "litellm_parent_otel_span": litellm_parent_otel_span,
     }
 
     return litellm_params