fix - send alert on router level exceptions

litellm/integrations/slack_alerting.py

@@ -1453,7 +1453,7 @@ Model Info:
             pass
         else:
             verbose_proxy_logger.debug(
-                "Error sending slack alert. Error=", response.text
+                "Error sending slack alert. Error={}".format(response.text)
             )
 
     async def async_log_success_event(self, kwargs, response_obj, start_time, end_time):

litellm/router.py

@@ -66,6 +66,7 @@ from litellm.types.llms.openai import (
 )
 from litellm.scheduler import Scheduler, FlowItem
 from typing import Iterable
+from litellm.router_utils.handle_error import send_llm_exception_alert
 
 
 class Router:
@@ -576,6 +577,14 @@ class Router:
 
             return response
         except Exception as e:
+            asyncio.create_task(
+                send_llm_exception_alert(
+                    litellm_router_instance=self,
+                    request_kwargs=kwargs,
+                    error_traceback_str=traceback.format_exc(),
+                    original_exception=e,
+                )
+            )
             raise e
 
     async def _acompletion(
@@ -4570,6 +4579,8 @@ class Router:
             default_webhook_url=router_alerting_config.webhook_url,
         )
 
+        self.slack_alerting_logger = _slack_alerting_logger
+
         litellm.callbacks.append(_slack_alerting_logger)
         litellm.success_callback.append(
             _slack_alerting_logger.response_taking_too_long_callback

litellm/router_utils/handle_error.py

@@ -0,0 +1,47 @@
+import asyncio
+import traceback
+from typing import TYPE_CHECKING, Any
+
+if TYPE_CHECKING:
+    from litellm.router import Router as _Router
+
+    LitellmRouter = _Router
+else:
+    LitellmRouter = Any
+
+
+async def send_llm_exception_alert(
+    litellm_router_instance: LitellmRouter,
+    request_kwargs: dict,
+    error_traceback_str: str,
+    original_exception,
+):
+    """
+    Sends a Slack / MS Teams alert for the LLM API call failure.
+
+    Parameters:
+        litellm_router_instance (_Router): The LitellmRouter instance.
+        original_exception (Any): The original exception that occurred.
+
+    Returns:
+        None
+    """
+    if litellm_router_instance.slack_alerting_logger is None:
+        return
+
+    if "proxy_server_request" in request_kwargs:
+        # Do not send any alert if it's a request from litellm proxy server request
+        # the proxy is already instrumented to send LLM API call failures
+        return
+
+    litellm_debug_info = getattr(original_exception, "litellm_debug_info", None)
+    exception_str = str(original_exception)
+    if litellm_debug_info is not None:
+        exception_str += litellm_debug_info
+    exception_str += f"\n\n{error_traceback_str[:2000]}"
+
+    await litellm_router_instance.slack_alerting_logger.send_alert(
+        message=f"LLM API call failed: `{exception_str}`",
+        level="High",
+        alert_type="llm_exceptions",
+    )