async def get_user_object(

fix

litellm/proxy/auth/auth_checks.py

@@ -386,8 +386,6 @@ async def get_user_object(
     - if valid, return LiteLLM_UserTable object with defined limits
     - if not, then raise an error
     """
-    if prisma_client is None:
-        raise Exception("No db connected")
 
     if user_id is None:
         return None
@@ -400,6 +398,8 @@ async def get_user_object(
         elif isinstance(cached_user_obj, LiteLLM_UserTable):
             return cached_user_obj
     # else, check db
+    if prisma_client is None:
+        raise Exception("No db connected")
     try:
 
         response = await prisma_client.db.litellm_usertable.find_unique(

litellm/proxy/hooks/parallel_request_limiter.py

@@ -322,10 +322,9 @@ class _PROXY_MaxParallelRequestsHandler(CustomLogger):
         # check if REQUEST ALLOWED for user_id
         user_id = user_api_key_dict.user_id
         if user_id is not None:
-            request_count_api_key = f"{user_id}::{precise_minute}::request_count"
-            _user_id_rate_limits = await self.internal_usage_cache.async_get_cache(
-                key=request_count_api_key,
-                litellm_parent_otel_span=user_api_key_dict.parent_otel_span,
+            _user_id_rate_limits = await self.get_internal_user_object(
+                user_id=user_id,
+                user_api_key_dict=user_api_key_dict,
             )
             # get user tpm/rpm limits
             if _user_id_rate_limits is not None and isinstance(
@@ -742,3 +741,25 @@ class _PROXY_MaxParallelRequestsHandler(CustomLogger):
                     str(e)
                 )
             )
+
+    async def get_internal_user_object(
+        self,
+        user_id: str,
+        user_api_key_dict: UserAPIKeyAuth,
+    ) -> Optional[dict]:
+        from litellm.proxy.auth.auth_checks import get_user_object
+        from litellm.proxy.proxy_server import prisma_client
+
+        _user_id_rate_limits = await get_user_object(
+            user_id=user_id,
+            prisma_client=prisma_client,
+            user_api_key_cache=self.internal_usage_cache.dual_cache,
+            user_id_upsert=False,
+            parent_otel_span=user_api_key_dict.parent_otel_span,
+            proxy_logging_obj=None,
+        )
+
+        if _user_id_rate_limits is None:
+            return None
+
+        return _user_id_rate_limits.model_dump()

litellm/tests/test_parallel_request_limiter.py

@@ -295,7 +295,13 @@ async def test_pre_call_hook_user_tpm_limits():
     local_cache = DualCache()
     # create user with tpm/rpm limits
     user_id = "test-user"
-    user_obj = {"tpm_limit": 9, "rpm_limit": 10}
+    user_obj = {
+        "tpm_limit": 9,
+        "rpm_limit": 10,
+        "user_id": user_id,
+        "user_email": "user_email",
+        "max_budget": None,
+    }
 
     local_cache.set_cache(key=user_id, value=user_obj)
 
@@ -331,6 +337,7 @@ async def test_pre_call_hook_user_tpm_limits():
     ## Expected cache val: {"current_requests": 0, "current_tpm": 0, "current_rpm": 1}
 
     try:
+        print("cache=local_cache", local_cache.in_memory_cache.cache_dict)
         await parallel_request_handler.async_pre_call_hook(
             user_api_key_dict=user_api_key_dict,
             cache=local_cache,