Merge pull request #42071 from BerriAI/litellm_remove_dead_telemetry_flag

.github/template.yaml

@@ -21,7 +21,7 @@ Parameters:
   WorkerConfigParameter:
     Type: String
     Description: Sample environment variable
-    Default: '{"model": null, "alias": null, "api_base": null, "api_version": "2023-07-01-preview", "debug": false, "temperature": null, "max_tokens": null, "request_timeout": 600, "max_budget": null, "telemetry": true, "drop_params": false, "add_function_to_prompt": false, "headers": null, "save": false, "config": null, "use_queue": false}'
+    Default: '{"model": null, "alias": null, "api_base": null, "api_version": "2023-07-01-preview", "debug": false, "temperature": null, "max_tokens": null, "request_timeout": 600, "max_budget": null, "drop_params": false, "add_function_to_prompt": false, "headers": null, "save": false, "config": null, "use_queue": false}'
 
 Resources:
   MyUrlFunctionPermissions:

cookbook/livekit_agent_sdk/config.example.yaml

@@ -15,7 +15,6 @@ model_list:
 
 litellm_settings:
   drop_params: True
-  telemetry: False
 
 general_settings:
   master_key: sk-1234  # Change this to a secure key

cookbook/misc/config.yaml

@@ -55,7 +55,6 @@ litellm_settings:
   # budget_duration: 30d
   num_retries: 5
   request_timeout: 600
-  telemetry: False
   context_window_fallbacks: [{"gpt-3.5-turbo": ["gpt-3.5-turbo-large"]}]
 
 general_settings: 

litellm/__init__.py

@@ -240,7 +240,6 @@ email: Optional[str] = (
 token: Optional[str] = (
     None  # Not used anymore, will be removed in next MAJOR release - https://github.com/BerriAI/litellm/discussions/648
 )
-telemetry = True
 max_tokens: int = DEFAULT_MAX_TOKENS  # OpenAI Defaults
 drop_params = drop_params_env_flag(os.environ, verbose_logger)
 modify_params = bool(os.getenv("LITELLM_MODIFY_PARAMS", False))

litellm/proxy/dev_config.yaml

@@ -212,7 +212,6 @@ sandbox_tools:
 
 litellm_settings:
   drop_params: True
-  telemetry: False
   code_interpreter_interception_params:
     enabled: true
     sandbox_tool_name: e2b_sandbox

litellm/proxy/example_config_yaml/oai_misc_config.yaml

@@ -38,7 +38,6 @@ litellm_settings:
   # budget_duration: 30d
   num_retries: 5
   request_timeout: 600
-  telemetry: False
   context_window_fallbacks: [{"gpt-5-mini": ["gpt-5.5"]}]
   default_team_settings: 
     - team_id: team-1

litellm/proxy/proxy_cli.py

@@ -56,8 +56,6 @@ if litellm_mode == "DEV":
     load_dotenv()
 from enum import Enum
 
-telemetry: Final = None
-
 
 class LiteLLMDatabaseConnectionPool(Enum):
     database_connection_pool_limit = 10
@@ -758,9 +756,11 @@ class ProxyInitializationHelpers:
 )
 @click.option(
     "--telemetry",
-    default=True,
+    default=None,
     type=bool,
-    help="Helps us know if people are using this feature. Turn this off by doing `--telemetry False`",
+    hidden=True,
+    expose_value=False,
+    help="Deprecated no-op kept so existing start commands still parse",
 )
 @click.option(
     "--log_config",
@@ -977,7 +977,6 @@ def run_server(
     add_function_to_prompt,
     config,
     max_budget,
-    telemetry,
     test,
     local,
     num_workers,
@@ -1082,7 +1081,6 @@ def run_server(
             max_tokens=max_tokens,
             request_timeout=request_timeout,
             max_budget=max_budget,
-            telemetry=telemetry,
             drop_params=drop_params,
             add_function_to_prompt=add_function_to_prompt,
             headers=headers,

litellm/proxy/proxy_server.py

@@ -1230,12 +1230,12 @@ async def proxy_startup_event(app: FastAPI) -> AsyncGenerator[None, None]:
                 general_settings,
             ) = await proxy_config.load_config(router=llm_router, config_file_path=worker_config)
         elif isinstance(worker_config, dict):
-            await initialize(**worker_config)
+            await initialize_from_worker_config(worker_config)
         else:
             # if not, assume it's a json string
             worker_config = json.loads(worker_config)
             if isinstance(worker_config, dict):
-                await initialize(**worker_config)
+                await initialize_from_worker_config(worker_config)
 
     enforce_master_key_boot_verdict(
         await with_stored_secrets_counted(
@@ -2426,7 +2426,6 @@ user_debug = False
 user_max_tokens = None
 user_request_timeout = None
 user_temperature = None
-user_telemetry = True
 user_config: Final = None
 user_headers = None
 user_config_file_path: str | None = None
@@ -8621,6 +8620,14 @@ def save_worker_config(**data):
     os.environ["WORKER_CONFIG"] = json.dumps(data)
 
 
+LEGACY_WORKER_CONFIG_KEYS: Final = frozenset({"telemetry"})
+
+
+async def initialize_from_worker_config(worker_config: Mapping[str, object]) -> None:
+    supported: Final = MappingProxyType({k: v for k, v in worker_config.items() if k not in LEGACY_WORKER_CONFIG_KEYS})
+    await initialize(**supported)
+
+
 async def initialize(
     model=None,
     alias=None,
@@ -8632,7 +8639,6 @@ async def initialize(
     max_tokens=None,
     request_timeout=600,
     max_budget=None,
-    telemetry=False,
     drop_params=True,
     add_function_to_prompt=True,
     headers=None,
@@ -8648,7 +8654,6 @@ async def initialize(
         user_user_max_tokens, \
         user_request_timeout, \
         user_temperature, \
-        user_telemetry, \
         user_headers, \
         experimental, \
         llm_model_list, \
@@ -8755,7 +8760,6 @@ async def initialize(
         dynamic_config["general"]["max_budget"] = litellm.max_budget
     if experimental:
         pass
-    user_telemetry = telemetry
 
 
 # for streaming

litellm/proxy/wildcard_config.yaml

@@ -49,4 +49,3 @@ general_settings:
 
 litellm_settings:
   drop_params: True
-  telemetry: False

proxy_server_config.yaml

@@ -173,7 +173,6 @@ litellm_settings:
   # budget_duration: 30d
   num_retries: 5
   request_timeout: 600
-  telemetry: False
   context_window_fallbacks: [{"gpt-3.5-turbo": ["gpt-3.5-turbo-large"]}]
   default_team_settings: 
     - team_id: team-1

scripts/benchmark_anthropic_messages_perf.py

@@ -256,7 +256,6 @@ general_settings:
   master_key: {api_key}
 
 litellm_settings:
-  telemetry: false
 """,
         encoding="utf-8",
     )

scripts/benchmark_chat_completions_perf.py

@@ -228,7 +228,6 @@ general_settings:
 
 litellm_settings:
   drop_params: true
-  telemetry: false
 """,
         encoding="utf-8",
     )

tests/integration/_support/process.py

@@ -74,8 +74,6 @@ def owned_proxy(gateway: Gateway, directory: Path, overrides: Mapping[str, str],
                 str(port),
                 "--num_workers",
                 "1",
-                "--telemetry",
-                "False",
                 "--use_prisma_db_push",
                 "--enforce_prisma_migration_check",
             ],

tests/logging_callback_tests/test_bedrock_knowledgebase_hook.py

@@ -883,7 +883,6 @@ async def test_provider_specific_fields_in_proxy_http_response(
         max_tokens=None,
         request_timeout=600,
         max_budget=None,
-        telemetry=False,
         drop_params=True,
         add_function_to_prompt=False,
         headers=None,

tests/proxy_unit_tests/test_proxy_utils.py

@@ -1112,7 +1112,6 @@ def test_settings_store_preserves_yaml_team_configuration_when_db_value_is_null(
         },
         "param_name": "litellm_settings",
         "db_param_value": {
-            "telemetry": False,
             "drop_params": True,
             "num_retries": 5,
             "request_timeout": 600,

tests/test_litellm/llms/wandb/test_wandb_chat_transformation.py

@@ -47,7 +47,6 @@ WANDB_REASONING_MODELS: Final = (
 @pytest.fixture
 def wandb_test_config(local_model_cost_map, monkeypatch: pytest.MonkeyPatch) -> None:
     monkeypatch.setattr(litellm, "disable_aiohttp_transport", True)
-    monkeypatch.setattr(litellm, "telemetry", False)
     monkeypatch.setattr(litellm, "drop_params", False)
 
 

tests/test_litellm/proxy/proxy_server/test_lifecycle.py

@@ -43,6 +43,7 @@ from litellm.proxy.proxy_server import (
     cost_tracking,
     get_litellm_model_info,
     initialize,
+    initialize_from_worker_config,
     load_from_azure_key_vault,
     proxy_shutdown_event,
     proxy_startup_event,
@@ -376,7 +377,7 @@ def _lit4152_worker_config_dict():
         "master_key": _LIT4152_SECRETS[0],
         "database_url": _LIT4152_SECRETS[3],
         "api_key": _LIT4152_SECRETS[2],
-        "telemetry": True,
+        "drop_params": True,
     }
 
 
@@ -394,7 +395,7 @@ def test__redact_worker_config_for_logging_dict_masks_all_secret_shapes():
         assert secret not in rendered, f"leak: {secret} in {rendered!r}"
     assert isinstance(redacted, dict)
     assert redacted["model"] == "openai/gpt-4o-mini"
-    assert redacted["telemetry"] is True
+    assert redacted["drop_params"] is True
 
 
 def test__redact_worker_config_for_logging_json_string_round_trips_masked():
@@ -501,7 +502,7 @@ def test__redact_worker_config_for_logging_masks_nested_secret_fields():
 def test_initialize_signature_is_async_with_expected_params():
     sig = inspect.signature(initialize)
     # Hard-coded so a signature change (param added/removed) trips the gate.
-    expected_param_count = 17
+    expected_param_count = 16
     observed = {
         "is_async": inspect.iscoroutinefunction(initialize),
         "param_count": len(sig.parameters),
@@ -522,6 +523,16 @@ async def test_initialize_invalid_unexpected_kwarg_raises_type_error():
         await initialize(this_is_not_a_real_kwarg=True)
 
 
+@pytest.mark.asyncio
+async def test_initialize_from_worker_config_drops_legacy_telemetry_key():
+    with pytest.raises(TypeError):
+        await initialize(telemetry=True)
+    await initialize_from_worker_config({"telemetry": True, "request_timeout": 77})
+    assert ps.user_request_timeout == 77
+    with pytest.raises(TypeError):
+        await initialize_from_worker_config({"this_is_not_a_real_kwarg": True})
+
+
 # ---------------------------------------------------------------------------
 # load_from_azure_key_vault
 # ---------------------------------------------------------------------------

tests/test_litellm/proxy/test_proxy_cli.py

@@ -617,6 +617,15 @@ class TestProxyInitializationHelpers:
             assert "Skipping server startup" in result.output
             mock_uvicorn_run.assert_not_called()
 
+            result = runner.invoke(
+                run_server, ["--local", "--skip_server_startup", "--telemetry", "False"]
+            )
+            assert (
+                result.exit_code == 0
+            ), f"exit_code={result.exit_code}, output={result.output}"
+            assert "Skipping server startup" in result.output
+            assert "telemetry" not in runner.invoke(run_server, ["--help"]).output
+
             # --- normal startup ---
             mock_uvicorn_run.reset_mock()