Use litellm logging convention

litellm/__init__.py

@@ -2,6 +2,7 @@
 import threading, requests
 from typing import Callable, List, Optional, Dict, Union, Any
 from litellm.caching import Cache
+from litellm._logging import set_verbose
 import httpx
 
 input_callback: List[Union[str, Callable]] = []
@@ -11,7 +12,6 @@ callbacks: List[Callable] = []
 _async_success_callback: List[Callable] = [] # internal variable - async custom callbacks are routed here. 
 pre_call_rules: List[Callable] = []
 post_call_rules: List[Callable] = []
-set_verbose = False
 email: Optional[
     str
 ] = None  # Not used anymore, will be removed in next MAJOR release - https://github.com/BerriAI/litellm/discussions/648

litellm/_logging.py

@@ -0,0 +1,5 @@
+set_verbose = False
+
+def print_verbose(print_statement):
+    if set_verbose:
+        print(print_statement)

litellm/health_check.py

@@ -4,7 +4,7 @@ from typing import Optional
 
 import litellm
 import logging
-from concurrent.futures import ThreadPoolExecutor
+from litellm._logging import print_verbose
 
 
 logger = logging.getLogger(__name__)
@@ -47,7 +47,7 @@ async def _perform_health_check(model_list: list):
         try:
             await litellm.acompletion(**model_params)
         except Exception as e:
-            logger.exception("Health check failed for model %s", model_params["model"])
+            print_verbose(f"Health check failed for model {model_params['model']}. Error: {e}")
             return False
         
         return True