chore(telemetry): code cleanup

# What does this PR do?


## Test Plan
# What does this PR do?


## Test Plan

llama_stack/apis/conversations/conversations.py

@@ -21,7 +21,7 @@ from llama_stack.apis.agents.openai_responses import (
     OpenAIResponseOutputMessageWebSearchToolCall,
 )
 from llama_stack.apis.version import LLAMA_STACK_API_V1
-from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
+from llama_stack.core.telemetry.trace_protocol import trace_protocol
 from llama_stack.schema_utils import json_schema_type, register_schema, webmethod
 
 Metadata = dict[str, str]

llama_stack/apis/files/files.py

@@ -12,7 +12,7 @@ from pydantic import BaseModel, Field
 
 from llama_stack.apis.common.responses import Order
 from llama_stack.apis.version import LLAMA_STACK_API_V1
-from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
+from llama_stack.core.telemetry.trace_protocol import trace_protocol
 from llama_stack.schema_utils import json_schema_type, webmethod
 
 

llama_stack/apis/inference/inference.py

@@ -23,6 +23,7 @@ from llama_stack.apis.common.responses import Order
 from llama_stack.apis.models import Model
 from llama_stack.apis.telemetry import MetricResponseMixin
 from llama_stack.apis.version import LLAMA_STACK_API_V1, LLAMA_STACK_API_V1ALPHA
+from llama_stack.core.telemetry.trace_protocol import trace_protocol
 from llama_stack.models.llama.datatypes import (
     BuiltinTool,
     StopReason,
@@ -30,7 +31,6 @@ from llama_stack.models.llama.datatypes import (
     ToolDefinition,
     ToolPromptFormat,
 )
-from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
 from llama_stack.schema_utils import json_schema_type, register_schema, webmethod
 
 register_schema(ToolCall)

llama_stack/apis/models/models.py

@@ -11,7 +11,7 @@ from pydantic import BaseModel, ConfigDict, Field, field_validator
 
 from llama_stack.apis.resource import Resource, ResourceType
 from llama_stack.apis.version import LLAMA_STACK_API_V1
-from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
+from llama_stack.core.telemetry.trace_protocol import trace_protocol
 from llama_stack.schema_utils import json_schema_type, webmethod
 
 

llama_stack/apis/prompts/prompts.py

@@ -11,7 +11,7 @@ from typing import Protocol, runtime_checkable
 from pydantic import BaseModel, Field, field_validator, model_validator
 
 from llama_stack.apis.version import LLAMA_STACK_API_V1
-from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
+from llama_stack.core.telemetry.trace_protocol import trace_protocol
 from llama_stack.schema_utils import json_schema_type, webmethod
 
 

llama_stack/apis/safety/safety.py

@@ -12,7 +12,7 @@ from pydantic import BaseModel, Field
 from llama_stack.apis.inference import OpenAIMessageParam
 from llama_stack.apis.shields import Shield
 from llama_stack.apis.version import LLAMA_STACK_API_V1
-from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
+from llama_stack.core.telemetry.trace_protocol import trace_protocol
 from llama_stack.schema_utils import json_schema_type, webmethod
 
 

llama_stack/apis/shields/shields.py

@@ -10,7 +10,7 @@ from pydantic import BaseModel
 
 from llama_stack.apis.resource import Resource, ResourceType
 from llama_stack.apis.version import LLAMA_STACK_API_V1
-from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
+from llama_stack.core.telemetry.trace_protocol import trace_protocol
 from llama_stack.schema_utils import json_schema_type, webmethod
 
 

llama_stack/apis/tools/rag_tool.py

@@ -12,7 +12,7 @@ from typing_extensions import runtime_checkable
 
 from llama_stack.apis.common.content_types import URL, InterleavedContent
 from llama_stack.apis.version import LLAMA_STACK_API_V1
-from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
+from llama_stack.core.telemetry.trace_protocol import trace_protocol
 from llama_stack.schema_utils import json_schema_type, register_schema, webmethod
 
 

llama_stack/apis/tools/tools.py

@@ -13,7 +13,7 @@ from typing_extensions import runtime_checkable
 from llama_stack.apis.common.content_types import URL, InterleavedContent
 from llama_stack.apis.resource import Resource, ResourceType
 from llama_stack.apis.version import LLAMA_STACK_API_V1
-from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
+from llama_stack.core.telemetry.trace_protocol import trace_protocol
 from llama_stack.schema_utils import json_schema_type, webmethod
 
 from .rag_tool import RAGToolRuntime

llama_stack/apis/vector_io/vector_io.py

@@ -17,7 +17,7 @@ from pydantic import BaseModel, Field
 from llama_stack.apis.inference import InterleavedContent
 from llama_stack.apis.vector_stores import VectorStore
 from llama_stack.apis.version import LLAMA_STACK_API_V1
-from llama_stack.providers.utils.telemetry.trace_protocol import trace_protocol
+from llama_stack.core.telemetry.trace_protocol import trace_protocol
 from llama_stack.providers.utils.vector_io.vector_utils import generate_chunk_id
 from llama_stack.schema_utils import json_schema_type, webmethod
 from llama_stack.strong_typing.schema import register_schema

llama_stack/cli/stack/run.py

@@ -15,10 +15,10 @@ import yaml
 
 from llama_stack.cli.stack.utils import ImageType
 from llama_stack.cli.subcommand import Subcommand
-from llama_stack.core.datatypes import LoggingConfig, StackRunConfig
+from llama_stack.core.datatypes import StackRunConfig
 from llama_stack.core.stack import cast_image_name_to_string, replace_env_vars
 from llama_stack.core.utils.config_resolution import Mode, resolve_config_or_distro
-from llama_stack.log import get_logger
+from llama_stack.log import LoggingConfig, get_logger
 
 REPO_ROOT = Path(__file__).parent.parent.parent.parent
 

llama_stack/core/datatypes.py

@@ -31,6 +31,7 @@ from llama_stack.core.storage.datatypes import (
     StorageBackendType,
     StorageConfig,
 )
+from llama_stack.log import LoggingConfig
 from llama_stack.providers.datatypes import Api, ProviderSpec
 
 LLAMA_STACK_BUILD_CONFIG_VERSION = 2
@@ -195,14 +196,6 @@ class TelemetryConfig(BaseModel):
     enabled: bool = Field(default=False, description="enable or disable telemetry")
 
 
-class LoggingConfig(BaseModel):
-    category_levels: dict[str, str] = Field(
-        default_factory=dict,
-        description="""
- Dictionary of different logging configurations for different portions (ex: core, server) of llama stack""",
-    )
-
-
 class OAuth2JWKSConfig(BaseModel):
     # The JWKS URI for collecting public keys
     uri: str

llama_stack/core/distribution.py

@@ -25,7 +25,7 @@ from llama_stack.providers.datatypes import (
 logger = get_logger(name=__name__, category="core")
 
 
-INTERNAL_APIS = {Api.inspect, Api.providers, Api.prompts, Api.conversations, Api.telemetry}
+INTERNAL_APIS = {Api.inspect, Api.providers, Api.prompts, Api.conversations}
 
 
 def stack_apis() -> list[Api]:

llama_stack/core/library_client.py

@@ -44,11 +44,12 @@ from llama_stack.core.stack import (
     get_stack_run_config_from_distro,
     replace_env_vars,
 )
+from llama_stack.core.telemetry import Telemetry
+from llama_stack.core.telemetry.tracing import CURRENT_TRACE_CONTEXT, end_trace, setup_logger, start_trace
 from llama_stack.core.utils.config import redact_sensitive_fields
 from llama_stack.core.utils.context import preserve_contexts_async_generator
 from llama_stack.core.utils.exec import in_notebook
 from llama_stack.log import get_logger, setup_logging
-from llama_stack.providers.utils.telemetry.tracing import CURRENT_TRACE_CONTEXT, end_trace, setup_logger, start_trace
 from llama_stack.strong_typing.inspection import is_unwrapped_body_param
 
 logger = get_logger(name=__name__, category="core")
@@ -293,8 +294,8 @@ class AsyncLlamaStackAsLibraryClient(AsyncLlamaStackClient):
             raise _e
 
         assert self.impls is not None
-        if Api.telemetry in self.impls:
-            setup_logger(self.impls[Api.telemetry])
+        if self.config.telemetry.enabled:
+            setup_logger(Telemetry())
 
         if not os.environ.get("PYTEST_CURRENT_TEST"):
             console = Console()

llama_stack/core/resolver.py

@@ -27,7 +27,6 @@ from llama_stack.apis.safety import Safety
 from llama_stack.apis.scoring import Scoring
 from llama_stack.apis.scoring_functions import ScoringFunctions
 from llama_stack.apis.shields import Shields
-from llama_stack.apis.telemetry import Telemetry
 from llama_stack.apis.tools import ToolGroups, ToolRuntime
 from llama_stack.apis.vector_io import VectorIO
 from llama_stack.apis.vector_stores import VectorStore
@@ -98,7 +97,7 @@ def api_protocol_map(external_apis: dict[Api, ExternalApiSpec] | None = None) ->
         Api.files: Files,
         Api.prompts: Prompts,
         Api.conversations: Conversations,
-        Api.telemetry: Telemetry,
+        # Api.telemetry: Telemetry,
     }
 
     if external_apis:
@@ -241,24 +240,6 @@ def validate_and_prepare_providers(
         key = api_str if api not in router_apis else f"inner-{api_str}"
         providers_with_specs[key] = specs
 
-    # TODO: remove this logic, telemetry should not have providers.
-    # if telemetry has been enabled in the config initialize our internal impl
-    # telemetry is not an external API so it SHOULD NOT be auto-routed.
-    if run_config.telemetry.enabled:
-        specs = {}
-        p = InlineProviderSpec(
-            api=Api.telemetry,
-            provider_type="inline::meta-reference",
-            pip_packages=[],
-            optional_api_dependencies=[Api.datasetio],
-            module="llama_stack.providers.inline.telemetry.meta_reference",
-            config_class="llama_stack.providers.inline.telemetry.meta_reference.config.TelemetryConfig",
-            description="Meta's reference implementation of telemetry and observability using OpenTelemetry.",
-        )
-        spec = ProviderWithSpec(spec=p, provider_type="inline::meta-reference", provider_id="meta-reference")
-        specs["meta-reference"] = spec
-        providers_with_specs["telemetry"] = specs
-
     return providers_with_specs
 
 

llama_stack/core/routers/__init__.py

@@ -72,14 +72,6 @@ async def get_auto_router_impl(
         raise ValueError(f"API {api.value} not found in router map")
 
     api_to_dep_impl = {}
-    if run_config.telemetry.enabled:
-        api_to_deps = {
-            "inference": {"telemetry": Api.telemetry},
-        }
-        for dep_name, dep_api in api_to_deps.get(api.value, {}).items():
-            if dep_api in deps:
-                api_to_dep_impl[dep_name] = deps[dep_api]
-
     # TODO: move pass configs to routers instead
     if api == Api.inference:
         inference_ref = run_config.storage.stores.inference
@@ -92,6 +84,7 @@ async def get_auto_router_impl(
         )
         await inference_store.initialize()
         api_to_dep_impl["store"] = inference_store
+        api_to_dep_impl["telemetry_enabled"] = run_config.telemetry.enabled
 
     elif api == Api.vector_io:
         api_to_dep_impl["vector_stores_config"] = run_config.vector_stores

llama_stack/core/routers/inference.py

@@ -53,13 +53,13 @@ from llama_stack.apis.inference.inference import (
     OpenAIChatCompletionContentPartTextParam,
 )
 from llama_stack.apis.models import Model, ModelType
-from llama_stack.apis.telemetry import MetricEvent, MetricInResponse, Telemetry
+from llama_stack.apis.telemetry import MetricEvent, MetricInResponse
+from llama_stack.core.telemetry.tracing import enqueue_event, get_current_span
 from llama_stack.log import get_logger
 from llama_stack.models.llama.llama3.chat_format import ChatFormat
 from llama_stack.models.llama.llama3.tokenizer import Tokenizer
 from llama_stack.providers.datatypes import HealthResponse, HealthStatus, RoutingTable
 from llama_stack.providers.utils.inference.inference_store import InferenceStore
-from llama_stack.providers.utils.telemetry.tracing import enqueue_event, get_current_span
 
 logger = get_logger(name=__name__, category="core::routers")
 
@@ -70,14 +70,14 @@ class InferenceRouter(Inference):
     def __init__(
         self,
         routing_table: RoutingTable,
-        telemetry: Telemetry | None = None,
         store: InferenceStore | None = None,
+        telemetry_enabled: bool = False,
     ) -> None:
         logger.debug("Initializing InferenceRouter")
         self.routing_table = routing_table
-        self.telemetry = telemetry
+        self.telemetry_enabled = telemetry_enabled
         self.store = store
-        if self.telemetry:
+        if self.telemetry_enabled:
             self.tokenizer = Tokenizer.get_instance()
             self.formatter = ChatFormat(self.tokenizer)
 
@@ -159,7 +159,7 @@ class InferenceRouter(Inference):
         model: Model,
     ) -> list[MetricInResponse]:
         metrics = self._construct_metrics(prompt_tokens, completion_tokens, total_tokens, model)
-        if self.telemetry:
+        if self.telemetry_enabled:
             for metric in metrics:
                 enqueue_event(metric)
         return [MetricInResponse(metric=metric.metric, value=metric.value) for metric in metrics]
@@ -223,7 +223,7 @@ class InferenceRouter(Inference):
             # that we do not return an AsyncIterator, our tests expect a stream of chunks we cannot intercept currently.
 
         response = await provider.openai_completion(params)
-        if self.telemetry:
+        if self.telemetry_enabled:
             metrics = self._construct_metrics(
                 prompt_tokens=response.usage.prompt_tokens,
                 completion_tokens=response.usage.completion_tokens,
@@ -285,7 +285,7 @@ class InferenceRouter(Inference):
         if self.store:
             asyncio.create_task(self.store.store_chat_completion(response, params.messages))
 
-        if self.telemetry:
+        if self.telemetry_enabled:
             metrics = self._construct_metrics(
                 prompt_tokens=response.usage.prompt_tokens,
                 completion_tokens=response.usage.completion_tokens,
@@ -393,7 +393,7 @@ class InferenceRouter(Inference):
             else:
                 if hasattr(chunk, "delta"):
                     completion_text += chunk.delta
-                if hasattr(chunk, "stop_reason") and chunk.stop_reason and self.telemetry:
+                if hasattr(chunk, "stop_reason") and chunk.stop_reason and self.telemetry_enabled:
                     complete = True
                     completion_tokens = await self._count_tokens(completion_text)
             # if we are done receiving tokens
@@ -401,7 +401,7 @@ class InferenceRouter(Inference):
                 total_tokens = (prompt_tokens or 0) + (completion_tokens or 0)
 
                 # Create a separate span for streaming completion metrics
-                if self.telemetry:
+                if self.telemetry_enabled:
                     # Log metrics in the new span context
                     completion_metrics = self._construct_metrics(
                         prompt_tokens=prompt_tokens,
@@ -450,7 +450,7 @@ class InferenceRouter(Inference):
         total_tokens = (prompt_tokens or 0) + (completion_tokens or 0)
 
         # Create a separate span for completion metrics
-        if self.telemetry:
+        if self.telemetry_enabled:
             # Log metrics in the new span context
             completion_metrics = self._construct_metrics(
                 prompt_tokens=prompt_tokens,
@@ -548,7 +548,7 @@ class InferenceRouter(Inference):
                         completion_text += "".join(choice_data["content_parts"])
 
                     # Add metrics to the chunk
-                    if self.telemetry and hasattr(chunk, "usage") and chunk.usage:
+                    if self.telemetry_enabled and hasattr(chunk, "usage") and chunk.usage:
                         metrics = self._construct_metrics(
                             prompt_tokens=chunk.usage.prompt_tokens,
                             completion_tokens=chunk.usage.completion_tokens,

llama_stack/core/server/server.py

@@ -36,7 +36,6 @@ from llama_stack.apis.common.responses import PaginatedResponse
 from llama_stack.core.access_control.access_control import AccessDeniedError
 from llama_stack.core.datatypes import (
     AuthenticationRequiredError,
-    LoggingConfig,
     StackRunConfig,
     process_cors_config,
 )
@@ -53,19 +52,13 @@ from llama_stack.core.stack import (
     cast_image_name_to_string,
     replace_env_vars,
 )
+from llama_stack.core.telemetry import Telemetry
+from llama_stack.core.telemetry.tracing import CURRENT_TRACE_CONTEXT, setup_logger
 from llama_stack.core.utils.config import redact_sensitive_fields
 from llama_stack.core.utils.config_resolution import Mode, resolve_config_or_distro
 from llama_stack.core.utils.context import preserve_contexts_async_generator
-from llama_stack.log import get_logger, setup_logging
+from llama_stack.log import LoggingConfig, get_logger, setup_logging
 from llama_stack.providers.datatypes import Api
-from llama_stack.providers.inline.telemetry.meta_reference.config import TelemetryConfig
-from llama_stack.providers.inline.telemetry.meta_reference.telemetry import (
-    TelemetryAdapter,
-)
-from llama_stack.providers.utils.telemetry.tracing import (
-    CURRENT_TRACE_CONTEXT,
-    setup_logger,
-)
 
 from .auth import AuthenticationMiddleware
 from .quota import QuotaMiddleware
@@ -451,9 +444,7 @@ def create_app() -> StackApp:
             app.add_middleware(CORSMiddleware, **cors_config.model_dump())
 
     if config.telemetry.enabled:
-        setup_logger(impls[Api.telemetry])
-    else:
-        setup_logger(TelemetryAdapter(TelemetryConfig(), {}))
+        setup_logger(Telemetry())
 
     # Load external APIs if configured
     external_apis = load_external_apis(config)

llama_stack/core/server/tracing.py

@@ -7,8 +7,8 @@ from aiohttp import hdrs
 
 from llama_stack.core.external import ExternalApiSpec
 from llama_stack.core.server.routes import find_matching_route, initialize_route_impls
+from llama_stack.core.telemetry.tracing import end_trace, start_trace
 from llama_stack.log import get_logger
-from llama_stack.providers.utils.telemetry.tracing import end_trace, start_trace
 
 logger = get_logger(name=__name__, category="core::server")
 

llama_stack/core/telemetry/__init__.py

@@ -0,0 +1,26 @@
+from .telemetry import Telemetry
+from .trace_protocol import serialize_value, trace_protocol
+from .tracing import (
+    CURRENT_TRACE_CONTEXT,
+    ROOT_SPAN_MARKERS,
+    end_trace,
+    enqueue_event,
+    get_current_span,
+    setup_logger,
+    span,
+    start_trace,
+)
+
+__all__ = [
+    "Telemetry",
+    "trace_protocol",
+    "serialize_value",
+    "CURRENT_TRACE_CONTEXT",
+    "ROOT_SPAN_MARKERS",
+    "end_trace",
+    "enqueue_event",
+    "get_current_span",
+    "setup_logger",
+    "span",
+    "start_trace",
+]

llama_stack/core/telemetry/telemetry.py

@@ -24,14 +24,13 @@ from llama_stack.apis.telemetry import (
     SpanStartPayload,
     SpanStatus,
     StructuredLogEvent,
-    Telemetry,
     UnstructuredLogEvent,
 )
-from llama_stack.core.datatypes import Api
+from llama_stack.apis.telemetry import (
+    Telemetry as TelemetryBase,
+)
+from llama_stack.core.telemetry.tracing import ROOT_SPAN_MARKERS
 from llama_stack.log import get_logger
-from llama_stack.providers.utils.telemetry.tracing import ROOT_SPAN_MARKERS
-
-from .config import TelemetryConfig
 
 _GLOBAL_STORAGE: dict[str, dict[str | int, Any]] = {
     "active_spans": {},
@@ -50,9 +49,8 @@ def is_tracing_enabled(tracer):
         return span.is_recording()
 
 
-class TelemetryAdapter(Telemetry):
-    def __init__(self, _config: TelemetryConfig, deps: dict[Api, Any]) -> None:
-        self.datasetio_api = deps.get(Api.datasetio)
+class Telemetry(TelemetryBase):
+    def __init__(self) -> None:
         self.meter = None
 
         global _TRACER_PROVIDER

llama_stack/core/telemetry/trace_protocol.py

@@ -77,7 +77,7 @@ def trace_protocol[T](cls: type[T]) -> type[T]:
 
         @wraps(method)
         async def async_gen_wrapper(self: Any, *args: Any, **kwargs: Any) -> AsyncGenerator:
-            from llama_stack.providers.utils.telemetry import tracing
+            from llama_stack.core.telemetry import tracing
 
             class_name, method_name, span_attributes = create_span_context(self, *args, **kwargs)
 
@@ -92,7 +92,7 @@ def trace_protocol[T](cls: type[T]) -> type[T]:
 
         @wraps(method)
         async def async_wrapper(self: Any, *args: Any, **kwargs: Any) -> Any:
-            from llama_stack.providers.utils.telemetry import tracing
+            from llama_stack.core.telemetry import tracing
 
             class_name, method_name, span_attributes = create_span_context(self, *args, **kwargs)
 
@@ -107,7 +107,7 @@ def trace_protocol[T](cls: type[T]) -> type[T]:
 
         @wraps(method)
         def sync_wrapper(self: Any, *args: Any, **kwargs: Any) -> Any:
-            from llama_stack.providers.utils.telemetry import tracing
+            from llama_stack.core.telemetry import tracing
 
             class_name, method_name, span_attributes = create_span_context(self, *args, **kwargs)
 

llama_stack/core/telemetry/tracing.py

@@ -28,8 +28,8 @@ from llama_stack.apis.telemetry import (
     Telemetry,
     UnstructuredLogEvent,
 )
+from llama_stack.core.telemetry.trace_protocol import serialize_value
 from llama_stack.log import get_logger
-from llama_stack.providers.utils.telemetry.trace_protocol import serialize_value
 
 logger = get_logger(__name__, category="core")
 

llama_stack/log.py

@@ -9,15 +9,23 @@ import os
 import re
 from logging.config import dictConfig  # allow-direct-logging
 
+from pydantic import BaseModel, Field
 from rich.console import Console
 from rich.errors import MarkupError
 from rich.logging import RichHandler
 
-from llama_stack.core.datatypes import LoggingConfig
-
 # Default log level
 DEFAULT_LOG_LEVEL = logging.INFO
 
+
+class LoggingConfig(BaseModel):
+    category_levels: dict[str, str] = Field(
+        default_factory=dict,
+        description="""
+Dictionary of different logging configurations for different portions (ex: core, server) of llama stack""",
+    )
+
+
 # Predefined categories
 CATEGORIES = [
     "core",

llama_stack/providers/inline/agents/meta_reference/agent_instance.py

@@ -67,6 +67,7 @@ from llama_stack.apis.safety import Safety
 from llama_stack.apis.tools import ToolGroups, ToolInvocationResult, ToolRuntime
 from llama_stack.apis.vector_io import VectorIO
 from llama_stack.core.datatypes import AccessRule
+from llama_stack.core.telemetry import tracing
 from llama_stack.log import get_logger
 from llama_stack.models.llama.datatypes import (
     BuiltinTool,
@@ -78,7 +79,6 @@ from llama_stack.providers.utils.inference.openai_compat import (
     convert_tooldef_to_openai_tool,
 )
 from llama_stack.providers.utils.kvstore import KVStore
-from llama_stack.providers.utils.telemetry import tracing
 
 from .persistence import AgentPersistence
 from .safety import SafetyException, ShieldRunnerMixin

llama_stack/providers/inline/agents/meta_reference/responses/streaming.py

@@ -65,9 +65,9 @@ from llama_stack.apis.inference import (
     OpenAIChoice,
     OpenAIMessageParam,
 )
+from llama_stack.core.telemetry import tracing
 from llama_stack.log import get_logger
 from llama_stack.providers.utils.inference.prompt_adapter import interleaved_content_as_str
-from llama_stack.providers.utils.telemetry import tracing
 
 from .types import ChatCompletionContext, ChatCompletionResult
 from .utils import (

llama_stack/providers/inline/agents/meta_reference/responses/tool_executor.py

@@ -37,8 +37,8 @@ from llama_stack.apis.inference import (
 )
 from llama_stack.apis.tools import ToolGroups, ToolInvocationResult, ToolRuntime
 from llama_stack.apis.vector_io import VectorIO
+from llama_stack.core.telemetry import tracing
 from llama_stack.log import get_logger
-from llama_stack.providers.utils.telemetry import tracing
 
 from .types import ChatCompletionContext, ToolExecutionResult
 

llama_stack/providers/inline/agents/meta_reference/safety.py

@@ -8,8 +8,8 @@ import asyncio
 
 from llama_stack.apis.inference import Message
 from llama_stack.apis.safety import Safety, SafetyViolation, ViolationLevel
+from llama_stack.core.telemetry import tracing
 from llama_stack.log import get_logger
-from llama_stack.providers.utils.telemetry import tracing
 
 log = get_logger(name=__name__, category="agents::meta_reference")
 

llama_stack/providers/inline/telemetry/__init__.py

@@ -1,5 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.

llama_stack/providers/inline/telemetry/meta_reference/__init__.py

@@ -1,21 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.
-
-from typing import Any
-
-from llama_stack.core.datatypes import Api
-
-from .config import TelemetryConfig, TelemetrySink
-
-__all__ = ["TelemetryConfig", "TelemetrySink"]
-
-
-async def get_provider_impl(config: TelemetryConfig, deps: dict[Api, Any]):
-    from .telemetry import TelemetryAdapter
-
-    impl = TelemetryAdapter(config, deps)
-    await impl.initialize()
-    return impl

llama_stack/providers/inline/telemetry/meta_reference/config.py

@@ -1,47 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.
-
-from enum import StrEnum
-from typing import Any
-
-from pydantic import BaseModel, Field, field_validator
-
-
-class TelemetrySink(StrEnum):
-    OTEL_TRACE = "otel_trace"
-    OTEL_METRIC = "otel_metric"
-    CONSOLE = "console"
-
-
-class TelemetryConfig(BaseModel):
-    otel_exporter_otlp_endpoint: str | None = Field(
-        default=None,
-        description="The OpenTelemetry collector endpoint URL (base URL for traces, metrics, and logs). If not set, the SDK will use OTEL_EXPORTER_OTLP_ENDPOINT environment variable.",
-    )
-    service_name: str = Field(
-        # service name is always the same, use zero-width space to avoid clutter
-        default="\u200b",
-        description="The service name to use for telemetry",
-    )
-    sinks: list[TelemetrySink] = Field(
-        default_factory=list,
-        description="List of telemetry sinks to enable (possible values: otel_trace, otel_metric, console)",
-    )
-
-    @field_validator("sinks", mode="before")
-    @classmethod
-    def validate_sinks(cls, v):
-        if isinstance(v, str):
-            return [TelemetrySink(sink.strip()) for sink in v.split(",")]
-        return v or []
-
-    @classmethod
-    def sample_run_config(cls, __distro_dir__: str) -> dict[str, Any]:
-        return {
-            "service_name": "${env.OTEL_SERVICE_NAME:=\u200b}",
-            "sinks": "${env.TELEMETRY_SINKS:=}",
-            "otel_exporter_otlp_endpoint": "${env.OTEL_EXPORTER_OTLP_ENDPOINT:=}",
-        }

llama_stack/providers/remote/inference/watsonx/watsonx.py

@@ -22,11 +22,11 @@ from llama_stack.apis.inference.inference import (
 )
 from llama_stack.apis.models import Model
 from llama_stack.apis.models.models import ModelType
+from llama_stack.core.telemetry.tracing import get_current_span
 from llama_stack.log import get_logger
 from llama_stack.providers.remote.inference.watsonx.config import WatsonXConfig
 from llama_stack.providers.utils.inference.litellm_openai_mixin import LiteLLMOpenAIMixin
 from llama_stack.providers.utils.inference.openai_compat import prepare_openai_completion_params
-from llama_stack.providers.utils.telemetry.tracing import get_current_span
 
 logger = get_logger(name=__name__, category="providers::remote::watsonx")
 

llama_stack/providers/utils/inference/litellm_openai_mixin.py

@@ -256,7 +256,7 @@ class LiteLLMOpenAIMixin(
         params: OpenAIChatCompletionRequestWithExtraBody,
     ) -> OpenAIChatCompletion | AsyncIterator[OpenAIChatCompletionChunk]:
         # Add usage tracking for streaming when telemetry is active
-        from llama_stack.providers.utils.telemetry.tracing import get_current_span
+        from llama_stack.core.telemetry.tracing import get_current_span
 
         stream_options = params.stream_options
         if params.stream and get_current_span() is not None:

llama_stack/providers/utils/telemetry/__init__.py

@@ -1,5 +0,0 @@
-# Copyright (c) Meta Platforms, Inc. and affiliates.
-# All rights reserved.
-#
-# This source code is licensed under the terms described in the LICENSE file in
-# the root directory of this source tree.

tests/integration/telemetry/conftest.py

@@ -23,7 +23,7 @@ from opentelemetry.sdk.trace import ReadableSpan, TracerProvider
 from opentelemetry.sdk.trace.export import SimpleSpanProcessor
 from opentelemetry.sdk.trace.export.in_memory_span_exporter import InMemorySpanExporter
 
-import llama_stack.providers.inline.telemetry.meta_reference.telemetry as telemetry_module
+import llama_stack.core.telemetry.telemetry as telemetry_module
 from llama_stack.testing.api_recorder import patch_httpx_for_test_id
 from tests.integration.fixtures.common import instantiate_llama_stack_client
 

tests/unit/distribution/test_distribution.py

@@ -196,8 +196,6 @@ class TestProviderRegistry:
             assert internal_api not in apis, f"Internal API {internal_api} should not be in providable_apis"
 
         for api in apis:
-            if api == Api.telemetry:
-                continue
             module_name = f"llama_stack.providers.registry.{api.name.lower()}"
             try:
                 importlib.import_module(module_name)