Merge branch 'main' into rag-metadata-support

llama_stack/apis/agents/agents.py

@@ -415,6 +415,7 @@ class Agents(Protocol):
         :returns: If stream=False, returns a Turn object.
                   If stream=True, returns an SSE event stream of AgentTurnResponseStreamChunk
         """
+        ...
 
     @webmethod(
         route="/agents/{agent_id}/session/{session_id}/turn/{turn_id}/resume",
@@ -606,3 +607,4 @@ class Agents(Protocol):
         :param model: The underlying LLM used for completions.
         :param previous_response_id: (Optional) if specified, the new response will be a continuation of the previous response. This can be used to easily fork-off new responses from existing responses.
         """
+        ...

llama_stack/apis/eval/eval.py

@@ -95,6 +95,7 @@ class Eval(Protocol):
         :param benchmark_config: The configuration for the benchmark.
         :return: The job that was created to run the evaluation.
         """
+        ...
 
     @webmethod(route="/eval/benchmarks/{benchmark_id}/evaluations", method="POST")
     async def evaluate_rows(
@@ -112,6 +113,7 @@ class Eval(Protocol):
         :param benchmark_config: The configuration for the benchmark.
         :return: EvaluateResponse object containing generations and scores
         """
+        ...
 
     @webmethod(route="/eval/benchmarks/{benchmark_id}/jobs/{job_id}", method="GET")
     async def job_status(self, benchmark_id: str, job_id: str) -> Job:
@@ -140,3 +142,4 @@ class Eval(Protocol):
         :param job_id: The ID of the job to get the result of.
         :return: The result of the job.
         """
+        ...

llama_stack/distribution/providers.py

@@ -99,7 +99,7 @@ class ProviderImpl(Providers):
             try:
                 health = await asyncio.wait_for(impl.health(), timeout=timeout)
                 return api_name, health
-            except asyncio.TimeoutError:
+            except (asyncio.TimeoutError, TimeoutError):
                 return (
                     api_name,
                     HealthResponse(

llama_stack/distribution/routers/routers.py

@@ -630,7 +630,7 @@ class InferenceRouter(Inference):
                     continue
                 health = await asyncio.wait_for(impl.health(), timeout=timeout)
                 health_statuses[provider_id] = health
-            except asyncio.TimeoutError:
+            except (asyncio.TimeoutError, TimeoutError):
                 health_statuses[provider_id] = HealthResponse(
                     status=HealthStatus.ERROR,
                     message=f"Health check timed out after {timeout} seconds",

llama_stack/distribution/server/server.py

@@ -114,7 +114,7 @@ def translate_exception(exc: Exception) -> HTTPException | RequestValidationErro
         return HTTPException(status_code=400, detail=str(exc))
     elif isinstance(exc, PermissionError):
         return HTTPException(status_code=403, detail=f"Permission denied: {str(exc)}")
-    elif isinstance(exc, TimeoutError):
+    elif isinstance(exc, asyncio.TimeoutError | TimeoutError):
         return HTTPException(status_code=504, detail=f"Operation timed out: {str(exc)}")
     elif isinstance(exc, NotImplementedError):
         return HTTPException(status_code=501, detail=f"Not implemented: {str(exc)}")
@@ -139,7 +139,7 @@ async def shutdown(app):
                 await asyncio.wait_for(impl.shutdown(), timeout=5)
             else:
                 logger.warning("No shutdown method for %s", impl_name)
-        except asyncio.TimeoutError:
+        except (asyncio.TimeoutError, TimeoutError):
             logger.exception("Shutdown timeout for %s ", impl_name, exc_info=True)
         except (Exception, asyncio.CancelledError) as e:
             logger.exception("Failed to shutdown %s: %s", impl_name, {e})

llama_stack/providers/inline/tool_runtime/rag/memory.py

@@ -106,7 +106,9 @@ class MemoryToolRuntimeImpl(ToolsProtocolPrivate, ToolRuntime, RAGToolRuntime):
         query_config: RAGQueryConfig | None = None,
     ) -> RAGQueryResult:
         if not vector_db_ids:
-            return RAGQueryResult(content=None)
+            raise ValueError(
+                "No vector DBs were provided to the knowledge search tool. Please provide at least one vector DB ID."
+            )
 
         query_config = query_config or RAGQueryConfig()
         query = await generate_rag_query(

llama_stack/providers/remote/vector_io/chroma/chroma.py

@@ -26,8 +26,7 @@ from .config import ChromaVectorIOConfig as RemoteChromaVectorIOConfig
 
 log = logging.getLogger(__name__)
 
-
-ChromaClientType = chromadb.AsyncHttpClient | chromadb.PersistentClient
+ChromaClientType = chromadb.api.AsyncClientAPI | chromadb.api.ClientAPI
 
 
 # this is a helper to allow us to use async and non-async chroma clients interchangeably