feat - set allowed fails policy

litellm/tests/test_router_retries.py

@@ -128,12 +128,17 @@ async def test_router_retries_errors(sync_mode, error_type):
     ["AuthenticationErrorRetries", "ContentPolicyViolationErrorRetries"],  #
 )
 async def test_router_retry_policy(error_type):
-    from litellm.router import RetryPolicy
+    from litellm.router import RetryPolicy, AllowedFailsPolicy
 
     retry_policy = RetryPolicy(
         ContentPolicyViolationErrorRetries=3, AuthenticationErrorRetries=0
     )
 
+    allowed_fails_policy = AllowedFailsPolicy(
+        ContentPolicyViolationErrorAllowedFails=1000,
+        RateLimitErrorAllowedFails=100,
+    )
+
     router = Router(
         model_list=[
             {
@@ -156,6 +161,7 @@ async def test_router_retry_policy(error_type):
             },
         ],
         retry_policy=retry_policy,
+        allowed_fails_policy=allowed_fails_policy,
     )
 
     customHandler = MyCustomHandler()

litellm/types/router.py

@@ -385,8 +385,8 @@ class RouterErrors(enum.Enum):
 
 class AllowedFailsPolicy(BaseModel):
     """
-    Use this to set a custom number of allowed_fails for each exception type before cooling down a deployment
-    If RateLimitErrorRetries = 3, then 3 retries will be made for RateLimitError
+    Use this to set a custom number of allowed fails/minute before cooling down a deployment
+    If `AuthenticationErrorAllowedFails = 1000`, then 1000 AuthenticationError will be allowed before cooling down a deployment
 
     Mapping of Exception type to allowed_fails for each exception
     https://docs.litellm.ai/docs/exception_mapping