fix(backend): Avoid long synchronous call to block FastAPI event-loop (#8429)

autogpt_platform/autogpt_libs/autogpt_libs/supabase_integration_credentials_store/store.py

@@ -6,7 +6,7 @@ if TYPE_CHECKING:
     from redis import Redis
     from backend.executor.database import DatabaseManager
 
-from autogpt_libs.utils.cache import thread_cached_property
+from autogpt_libs.utils.cache import thread_cached
 from autogpt_libs.utils.synchronize import RedisKeyedMutex
 
 from .types import (
@@ -21,8 +21,9 @@ from .types import (
 class SupabaseIntegrationCredentialsStore:
     def __init__(self, redis: "Redis"):
         self.locks = RedisKeyedMutex(redis)
-        
-    @thread_cached_property
+
+    @property
+    @thread_cached
     def db_manager(self) -> "DatabaseManager":
         from backend.executor.database import DatabaseManager
         from backend.util.service import get_service_client

autogpt_platform/autogpt_libs/autogpt_libs/utils/cache.py

@@ -1,8 +1,6 @@
+from typing import Callable, TypeVar, ParamSpec
 import threading
-from functools import wraps
-from typing import Callable, ParamSpec, TypeVar
 
-T = TypeVar("T")
 P = ParamSpec("P")
 R = TypeVar("R")
 
@@ -10,7 +8,6 @@ R = TypeVar("R")
 def thread_cached(func: Callable[P, R]) -> Callable[P, R]:
     thread_local = threading.local()
 
-    @wraps(func)
     def wrapper(*args: P.args, **kwargs: P.kwargs) -> R:
         cache = getattr(thread_local, "cache", None)
         if cache is None:
@@ -21,7 +18,3 @@ def thread_cached(func: Callable[P, R]) -> Callable[P, R]:
         return cache[key]
 
     return wrapper
-
-
-def thread_cached_property(func: Callable[[T], R]) -> property:
-    return property(thread_cached(func))

autogpt_platform/backend/backend/executor/scheduler.py

@@ -4,7 +4,7 @@ from datetime import datetime
 
 from apscheduler.schedulers.background import BackgroundScheduler
 from apscheduler.triggers.cron import CronTrigger
-from autogpt_libs.utils.cache import thread_cached_property
+from autogpt_libs.utils.cache import thread_cached
 
 from backend.data.block import BlockInput
 from backend.data.schedule import (
@@ -37,7 +37,8 @@ class ExecutionScheduler(AppService):
     def get_port(cls) -> int:
         return Config().execution_scheduler_port
 
-    @thread_cached_property
+    @property
+    @thread_cached
     def execution_client(self) -> ExecutionManager:
         return get_service_client(ExecutionManager)
 

autogpt_platform/backend/backend/server/integrations/router.py

@@ -29,7 +29,7 @@ class LoginResponse(BaseModel):
 
 
 @router.get("/{provider}/login")
-async def login(
+def login(
     provider: Annotated[str, Path(title="The provider to initiate an OAuth flow for")],
     user_id: Annotated[str, Depends(get_user_id)],
     request: Request,
@@ -60,7 +60,7 @@ class CredentialsMetaResponse(BaseModel):
 
 
 @router.post("/{provider}/callback")
-async def callback(
+def callback(
     provider: Annotated[str, Path(title="The target provider for this OAuth exchange")],
     code: Annotated[str, Body(title="Authorization code acquired by user login")],
     state_token: Annotated[str, Body(title="Anti-CSRF nonce")],
@@ -115,7 +115,7 @@ async def callback(
 
 
 @router.get("/{provider}/credentials")
-async def list_credentials(
+def list_credentials(
     provider: Annotated[str, Path(title="The provider to list credentials for")],
     user_id: Annotated[str, Depends(get_user_id)],
 ) -> list[CredentialsMetaResponse]:
@@ -133,7 +133,7 @@ async def list_credentials(
 
 
 @router.get("/{provider}/credentials/{cred_id}")
-async def get_credential(
+def get_credential(
     provider: Annotated[str, Path(title="The provider to retrieve credentials for")],
     cred_id: Annotated[str, Path(title="The ID of the credentials to retrieve")],
     user_id: Annotated[str, Depends(get_user_id)],
@@ -149,7 +149,7 @@ async def get_credential(
 
 
 @router.post("/{provider}/credentials", status_code=201)
-async def create_api_key_credentials(
+def create_api_key_credentials(
     user_id: Annotated[str, Depends(get_user_id)],
     provider: Annotated[str, Path(title="The provider to create credentials for")],
     api_key: Annotated[str, Body(title="The API key to store")],
@@ -184,7 +184,7 @@ class CredentialsDeletionResponse(BaseModel):
 
 
 @router.delete("/{provider}/credentials/{cred_id}")
-async def delete_credentials(
+def delete_credentials(
     request: Request,
     provider: Annotated[str, Path(title="The provider to delete credentials for")],
     cred_id: Annotated[str, Path(title="The ID of the credentials to delete")],

autogpt_platform/backend/backend/server/rest_api.py

@@ -1,3 +1,4 @@
+import asyncio
 import inspect
 import logging
 from collections import defaultdict
@@ -7,7 +8,7 @@ from typing import Annotated, Any, Dict
 
 import uvicorn
 from autogpt_libs.auth.middleware import auth_middleware
-from autogpt_libs.utils.cache import thread_cached_property
+from autogpt_libs.utils.cache import thread_cached
 from fastapi import APIRouter, Body, Depends, FastAPI, HTTPException, Request
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.responses import JSONResponse
@@ -307,11 +308,13 @@ class AgentServer(AppService):
 
         return wrapper
 
-    @thread_cached_property
+    @property
+    @thread_cached
     def execution_manager_client(self) -> ExecutionManager:
         return get_service_client(ExecutionManager)
 
-    @thread_cached_property
+    @property
+    @thread_cached
     def execution_scheduler_client(self) -> ExecutionScheduler:
         return get_service_client(ExecutionScheduler)
 
@@ -516,7 +519,7 @@ class AgentServer(AppService):
             user_id=user_id,
         )
 
-    async def execute_graph(
+    def execute_graph(
         self,
         graph_id: str,
         node_input: dict[Any, Any],
@@ -539,7 +542,9 @@ class AgentServer(AppService):
                 404, detail=f"Agent execution #{graph_exec_id} not found"
             )
 
-        self.execution_manager_client.cancel_execution(graph_exec_id)
+        await asyncio.to_thread(
+            lambda: self.execution_manager_client.cancel_execution(graph_exec_id)
+        )
 
         # Retrieve & return canceled graph execution in its final state
         return await execution_db.get_execution_results(graph_exec_id)
@@ -614,10 +619,16 @@ class AgentServer(AppService):
         graph = await graph_db.get_graph(graph_id, user_id=user_id)
         if not graph:
             raise HTTPException(status_code=404, detail=f"Graph #{graph_id} not found.")
-        execution_scheduler = self.execution_scheduler_client
+
         return {
-            "id": execution_scheduler.add_execution_schedule(
-                graph_id, graph.version, cron, input_data, user_id=user_id
+            "id": await asyncio.to_thread(
+                lambda: self.execution_scheduler_client.add_execution_schedule(
+                    graph_id=graph_id,
+                    graph_version=graph.version,
+                    cron=cron,
+                    input_data=input_data,
+                    user_id=user_id,
+                )
             )
         }
 

autogpt_platform/backend/backend/usecases/block_autogen.py

@@ -252,7 +252,7 @@ async def block_autogen_agent():
         test_user = await create_test_user()
         test_graph = await create_graph(create_test_graph(), user_id=test_user.id)
         input_data = {"input": "Write me a block that writes a string into a file."}
-        response = await server.agent_server.execute_graph(
+        response = server.agent_server.execute_graph(
             test_graph.id, input_data, test_user.id
         )
         print(response)

autogpt_platform/backend/backend/usecases/reddit_marketing.py

@@ -156,7 +156,7 @@ async def reddit_marketing_agent():
         test_user = await create_test_user()
         test_graph = await create_graph(create_test_graph(), user_id=test_user.id)
         input_data = {"subreddit": "AutoGPT"}
-        response = await server.agent_server.execute_graph(
+        response = server.agent_server.execute_graph(
             test_graph.id, input_data, test_user.id
         )
         print(response)

autogpt_platform/backend/backend/usecases/sample.py

@@ -78,7 +78,7 @@ async def sample_agent():
         test_user = await create_test_user()
         test_graph = await create_graph(create_test_graph(), test_user.id)
         input_data = {"input_1": "Hello", "input_2": "World"}
-        response = await server.agent_server.execute_graph(
+        response = server.agent_server.execute_graph(
             test_graph.id, input_data, test_user.id
         )
         print(response)

autogpt_platform/backend/test/executor/test_manager.py

@@ -22,7 +22,7 @@ async def execute_graph(
     num_execs: int = 4,
 ) -> str:
     # --- Test adding new executions --- #
-    response = await agent_server.execute_graph(test_graph.id, input_data, test_user.id)
+    response = agent_server.execute_graph(test_graph.id, input_data, test_user.id)
     graph_exec_id = response["id"]
 
     # Execution queue should be empty