test(test_router_fallbacks.py): update to use async logging

litellm/tests/langfuse.log

@@ -1,5 +1,5 @@
 Task exception was never retrieved
-future: <Task finished name='Task-295' coro=<QueryEngine.aclose() done, defined at /opt/homebrew/lib/python3.11/site-packages/prisma/engine/query.py:110> exception=RuntimeError('Event loop is closed')>
+future: <Task finished name='Task-299' coro=<QueryEngine.aclose() done, defined at /opt/homebrew/lib/python3.11/site-packages/prisma/engine/query.py:110> exception=RuntimeError('Event loop is closed')>
 Traceback (most recent call last):
   File "/opt/homebrew/lib/python3.11/site-packages/prisma/engine/query.py", line 112, in aclose
     await self._close_session()
@@ -29,3 +29,5 @@ Traceback (most recent call last):
     raise RuntimeError('Event loop is closed')
 RuntimeError: Event loop is closed
 Giving up get_data(...) after 3 tries (prisma.errors.ClientNotConnectedError: Client is not connected to the query engine, you must call `connect()` before attempting to query data.)
+Giving up get_data(...) after 3 tries (prisma.errors.ClientNotConnectedError: Client is not connected to the query engine, you must call `connect()` before attempting to query data.)
+Giving up get_data(...) after 3 tries (prisma.errors.ClientNotConnectedError: Client is not connected to the query engine, you must call `connect()` before attempting to query data.)

litellm/tests/test_configs/test_config_no_auth.yaml

@@ -43,3 +43,15 @@ model_list:
     description: this is a test openai model
     id: 79fc75bf-8e1b-47d5-8d24-9365a854af03
   model_name: test_openai_models
+- litellm_params:
+    model: gpt-3.5-turbo
+  model_info:
+    description: this is a test openai model
+    id: 78ec9388-427e-49c7-bdf2-355f00a6b246
+  model_name: test_openai_models
+- litellm_params:
+    model: gpt-3.5-turbo
+  model_info:
+    description: this is a test openai model
+    id: cb373690-360e-4aba-8fae-71de12f5278e
+  model_name: test_openai_models

litellm/tests/test_router_fallbacks.py

@@ -32,6 +32,12 @@ class MyCustomHandler(CustomLogger):
         print(f"self.previous_models: {self.previous_models}")
         print(f"On Success")
     
+    async def async_log_success_event(self, kwargs, response_obj, start_time, end_time): 
+        print(f"previous_models: {kwargs['litellm_params']['metadata']['previous_models']}")
+        self.previous_models += len(kwargs["litellm_params"]["metadata"]["previous_models"]) # {"previous_models": [{"model": litellm_model_name, "exception_type": AuthenticationError, "exception_string": <complete_traceback>}]}
+        print(f"self.previous_models: {self.previous_models}")
+        print(f"On Success")
+
     def log_failure_event(self, kwargs, response_obj, start_time, end_time): 
         print(f"On Failure")
 
@@ -111,13 +117,13 @@ def test_sync_fallbacks():
         print(e)
 # test_sync_fallbacks() 
 
-def test_async_fallbacks(): 
+@pytest.mark.asyncio
+async def test_async_fallbacks(): 
     litellm.set_verbose = False
     router = Router(model_list=model_list, 
                 fallbacks=[{"azure/gpt-3.5-turbo": ["gpt-3.5-turbo"]}], 
                 context_window_fallbacks=[{"azure/gpt-3.5-turbo-context-fallback": ["gpt-3.5-turbo-16k"]}, {"gpt-3.5-turbo": ["gpt-3.5-turbo-16k"]}],
                 set_verbose=False)
-    async def test_get_response():
     customHandler = MyCustomHandler()
     litellm.callbacks = [customHandler]
     user_message = "Hello, how are you?"
@@ -125,7 +131,7 @@ def test_async_fallbacks():
     try:
         response = await router.acompletion(**kwargs)
         print(f"customHandler.previous_models: {customHandler.previous_models}")
-            time.sleep(0.05) # allow a delay as success_callbacks are on a separate thread
+        await asyncio.sleep(0.05) # allow a delay as success_callbacks are on a separate thread
         assert customHandler.previous_models == 1 # 0 retries, 1 fallback
         router.reset()
     except litellm.Timeout as e: 
@@ -134,7 +140,6 @@ def test_async_fallbacks():
         pytest.fail(f"An exception occurred: {e}")
     finally:
         router.reset()
-    asyncio.run(test_get_response())
 
 # test_async_fallbacks()
 
@@ -184,11 +189,11 @@ def test_dynamic_fallbacks_sync():
 
 # test_dynamic_fallbacks_sync()
 
-def test_dynamic_fallbacks_async(): 
+@pytest.mark.asyncio
+async def test_dynamic_fallbacks_async(): 
     """
     Allow setting the fallback in the router.completion() call. 
     """
-    async def test_get_response():
     try: 
         customHandler = MyCustomHandler()
         litellm.callbacks = [customHandler]
@@ -199,11 +204,9 @@ def test_dynamic_fallbacks_async():
         kwargs["fallbacks"] = [{"azure/gpt-3.5-turbo": ["gpt-3.5-turbo"]}]
         response = await router.acompletion(**kwargs)
         print(f"response: {response}")
-            time.sleep(0.05) # allow a delay as success_callbacks are on a separate thread
+        await asyncio.sleep(0.05) # allow a delay as success_callbacks are on a separate thread
         assert customHandler.previous_models == 1 # 0 retries, 1 fallback
         router.reset()
     except Exception as e:
         pytest.fail(f"An exception occurred - {e}")
-    asyncio.run(test_get_response())
-
 # test_dynamic_fallbacks_async()