From f0fb8bdf456d8cef6edb68901f8139710d9c8ed9 Mon Sep 17 00:00:00 2001
From: Krrish Dholakia <krrishdholakia@gmail.com>
Date: Wed, 28 Aug 2024 16:38:05 -0700
Subject: [PATCH] fix(router.py): fix cooldown check

---
 litellm/proxy/_new_secret_config.yaml | 16 +++-------------
 litellm/router.py                     | 23 ++++++-----------------
 litellm/types/router.py               | 14 ++++++++++++++
 3 files changed, 23 insertions(+), 30 deletions(-)

diff --git a/litellm/proxy/_new_secret_config.yaml b/litellm/proxy/_new_secret_config.yaml
index e4609ce75e..d683983504 100644
--- a/litellm/proxy/_new_secret_config.yaml
+++ b/litellm/proxy/_new_secret_config.yaml
@@ -1,19 +1,9 @@
 model_list:
   - model_name: fake-openai-endpoint
     litellm_params:
-      model: gpt-3.5-turbo
-      # model: sagemaker/jumpstart-dft-hf-textgeneration1-mp-20240815-185614
-      # sagemaker_base_url: https://exampleopenaiendpoint-production.up.railway.app/invocations/
-      # api_base: https://exampleopenaiendpoint-production.up.railway.app
-      input_cost_per_token: 10
-      output_cost_per_token: 10
+      model: openai/my-fake-model
+      api_key: my-fake-key
+      api_base: https://exampleopenaiendpoint-production.up.railway.app/
 
-litellm_settings:
-  max_internal_user_budget: 0.00001
-  internal_user_budget_duration: "3s" # reset every 3seconds
-
-general_settings:
-  proxy_budget_rescheduler_min_time: 1
-  proxy_budget_rescheduler_max_time: 2
 
 
diff --git a/litellm/router.py b/litellm/router.py
index bc23972b63..37ca435d7c 100644
--- a/litellm/router.py
+++ b/litellm/router.py
@@ -92,6 +92,7 @@ from litellm.types.router import (
     RouterErrors,
     RouterGeneralSettings,
     RouterRateLimitError,
+    RouterRateLimitErrorBasic,
     updateDeployment,
     updateLiteLLMParams,
 )
@@ -4459,16 +4460,8 @@ class Router:
             """
 
             if _rate_limit_error is True:  # allow generic fallback logic to take place
-                model_ids = self.get_model_ids(model_name=model)
-                cooldown_time = self.cooldown_cache.get_min_cooldown(
-                    model_ids=model_ids
-                )
-                cooldown_list = self._get_cooldown_deployments()
-                raise RouterRateLimitError(
+                raise RouterRateLimitErrorBasic(
                     model=model,
-                    cooldown_time=cooldown_time,
-                    enable_pre_call_checks=True,
-                    cooldown_list=cooldown_list,
                 )
 
             elif _context_window_error is True:
@@ -4579,14 +4572,10 @@ class Router:
         litellm.print_verbose(f"initial list of deployments: {healthy_deployments}")
 
         if len(healthy_deployments) == 0:
-            model_ids = self.get_model_ids(model_name=model)
-            _cooldown_time = self.cooldown_cache.get_min_cooldown(model_ids=model_ids)
-            _cooldown_list = self._get_cooldown_deployments()
-            raise RouterRateLimitError(
-                model=model,
-                cooldown_time=_cooldown_time,
-                enable_pre_call_checks=self.enable_pre_call_checks,
-                cooldown_list=_cooldown_list,
+            raise ValueError(
+                "{}. You passed in model={}. There is no 'model_name' with this string ".format(
+                    RouterErrors.no_deployments_available.value, model
+                )
             )
 
         if litellm.model_alias_map and model in litellm.model_alias_map:
diff --git a/litellm/types/router.py b/litellm/types/router.py
index e38fec0758..b05e9be4a6 100644
--- a/litellm/types/router.py
+++ b/litellm/types/router.py
@@ -551,6 +551,20 @@ class RouterGeneralSettings(BaseModel):
     )  # if passed a model not llm_router model list, pass through the request to litellm.acompletion/embedding
 
 
+class RouterRateLimitErrorBasic(ValueError):
+    """
+    Raise a basic error inside helper functions.
+    """
+
+    def __init__(
+        self,
+        model: str,
+    ):
+        self.model = model
+        _message = f"{RouterErrors.no_deployments_available.value}."
+        super().__init__(_message)
+
+
 class RouterRateLimitError(ValueError):
     def __init__(
         self,