fix trace starting in library client (#655)

# What does this PR do?

Because of the way library client sets up async io boundaries, tracing
was broken with streaming. This PR fixes the tracing to start at the
right way to caputre the life time of async gen functions correctly.

Test plan:
Script ran:
https://gist.github.com/yanxi0830/f6645129e55ab12de3cd6ec71564c69e

Before: No spans returned for a session


Now: We see spans
<img width="1678" alt="Screenshot 2024-12-18 at 9 50 46 PM"
src="https://github.com/user-attachments/assets/58a3b0dd-a41c-489a-b89a-075e698a2c03"
/>

llama_stack/distribution/library_client.py

@@ -67,6 +67,7 @@ def in_notebook():
 def stream_across_asyncio_run_boundary(
     async_gen_maker,
     pool_executor: ThreadPoolExecutor,
+    path: Optional[str] = None,
 ) -> Generator[T, None, None]:
     result_queue = queue.Queue()
     stop_event = threading.Event()
@@ -74,6 +75,7 @@ def stream_across_asyncio_run_boundary(
     async def consumer():
         # make sure we make the generator in the event loop context
         gen = await async_gen_maker()
+        await start_trace(path, {"__location__": "library_client"})
         try:
             async for item in await gen:
                 result_queue.put(item)
@@ -85,6 +87,7 @@ def stream_across_asyncio_run_boundary(
         finally:
             result_queue.put(StopIteration)
             stop_event.set()
+            await end_trace()
 
     def run_async():
         # Run our own loop to avoid double async generator cleanup which is done
@@ -186,14 +189,34 @@ class LlamaStackAsLibraryClient(LlamaStackClient):
 
         return asyncio.run(self.async_client.initialize())
 
+    def _get_path(
+        self,
+        cast_to: Any,
+        options: Any,
+        *,
+        stream=False,
+        stream_cls=None,
+    ):
+        return options.url
+
     def request(self, *args, **kwargs):
+        path = self._get_path(*args, **kwargs)
         if kwargs.get("stream"):
             return stream_across_asyncio_run_boundary(
                 lambda: self.async_client.request(*args, **kwargs),
                 self.pool_executor,
+                path=path,
             )
         else:
-            return asyncio.run(self.async_client.request(*args, **kwargs))
+
+            async def _traced_request():
+                await start_trace(path, {"__location__": "library_client"})
+                try:
+                    return await self.async_client.request(*args, **kwargs)
+                finally:
+                    await end_trace()
+
+            return asyncio.run(_traced_request())
 
 
 class AsyncLlamaStackAsLibraryClient(AsyncLlamaStackClient):
@@ -206,7 +229,10 @@ class AsyncLlamaStackAsLibraryClient(AsyncLlamaStackClient):
 
         # when using the library client, we should not log to console since many
         # of our logs are intended for server-side usage
-        os.environ["TELEMETRY_SINKS"] = "sqlite"
+        current_sinks = os.environ.get("TELEMETRY_SINKS", "sqlite").split(",")
+        os.environ["TELEMETRY_SINKS"] = ",".join(
+            sink for sink in current_sinks if sink != "console"
+        )
 
         if config_path_or_template_name.endswith(".yaml"):
             config_path = Path(config_path_or_template_name)
@@ -295,8 +321,6 @@ class AsyncLlamaStackAsLibraryClient(AsyncLlamaStackClient):
 
         body = options.params or {}
         body |= options.json_data or {}
-        await start_trace(path, {"__location__": "library_client"})
-        try:
         func = self.endpoint_impls.get(path)
         if not func:
             raise ValueError(f"No endpoint found for {path}")
@@ -328,8 +352,6 @@ class AsyncLlamaStackAsLibraryClient(AsyncLlamaStackClient):
             stream_cls=None,
         )
         return response.parse()
-        finally:
-            await end_trace()
 
     async def _call_streaming(
         self,
@@ -341,8 +363,6 @@ class AsyncLlamaStackAsLibraryClient(AsyncLlamaStackClient):
         path = options.url
         body = options.params or {}
         body |= options.json_data or {}
-        await start_trace(path, {"__location__": "library_client"})
-        try:
         func = self.endpoint_impls.get(path)
         if not func:
             raise ValueError(f"No endpoint found for {path}")
@@ -384,8 +404,6 @@ class AsyncLlamaStackAsLibraryClient(AsyncLlamaStackClient):
             stream_cls=stream_cls,
         )
         return await response.parse()
-        finally:
-            await end_trace()
 
     def _convert_body(self, path: str, body: Optional[dict] = None) -> dict:
         if not body: