feat(router): add health-check-driven routing behind opt-in flag

Background health checks now feed deployment health state into the
router candidate-filtering pipeline. Unhealthy deployments are excluded
proactively instead of waiting for request failures to trigger cooldown.

Gated by `enable_health_check_routing: true` in general_settings.
Off by default — zero behavior change for existing users.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

litellm/proxy/health_check.py

@@ -210,20 +210,20 @@ async def _perform_health_check(
         _model_id = (model.get("model_info") or {}).get("id")
 
         if isinstance(is_healthy, dict) and "error" not in is_healthy:
-            cleaned = _clean_endpoint_data({**litellm_params, **is_healthy}, details)
+            endpoint_data = {**litellm_params, **is_healthy}
             if _model_id:
-                cleaned["model_id"] = _model_id
-            healthy_endpoints.append(cleaned)
+                endpoint_data["model_id"] = _model_id
+            healthy_endpoints.append(_clean_endpoint_data(endpoint_data, details))
         elif isinstance(is_healthy, dict):
-            cleaned = _clean_endpoint_data({**litellm_params, **is_healthy}, details)
+            endpoint_data = {**litellm_params, **is_healthy}
             if _model_id:
-                cleaned["model_id"] = _model_id
-            unhealthy_endpoints.append(cleaned)
+                endpoint_data["model_id"] = _model_id
+            unhealthy_endpoints.append(_clean_endpoint_data(endpoint_data, details))
         else:
-            cleaned = _clean_endpoint_data(litellm_params, details)
+            endpoint_data = {**litellm_params}
             if _model_id:
-                cleaned["model_id"] = _model_id
-            unhealthy_endpoints.append(cleaned)
+                endpoint_data["model_id"] = _model_id
+            unhealthy_endpoints.append(_clean_endpoint_data(endpoint_data, details))
 
     return healthy_endpoints, unhealthy_endpoints
 

litellm/proxy/proxy_server.py

@@ -37,7 +37,7 @@ import websockets
 import websockets.exceptions
 from pydantic import BaseModel, Json
 
-from litellm._uuid import uuid
+from litellm._litellm_uuid import uuid
 from litellm.constants import (
     AIOHTTP_CONNECTOR_LIMIT,
     AIOHTTP_CONNECTOR_LIMIT_PER_HOST,
@@ -2138,7 +2138,7 @@ def _write_health_state_to_router_cache(
                 sum(1 for s in states.values() if not s.get("is_healthy")),
             )
     except Exception as e:
-        verbose_proxy_logger.warning(
+        verbose_proxy_logger.debug(
             "Failed to write health state to router cache: %s", str(e)
         )
 

litellm/router.py

@@ -46,8 +46,8 @@ import litellm
 import litellm.litellm_core_utils
 import litellm.litellm_core_utils.exception_mapping_utils
 from litellm import get_secret_str
+from litellm._litellm_uuid import uuid
 from litellm._logging import verbose_router_logger
-from litellm._uuid import uuid
 from litellm.caching.caching import (
     DualCache,
     InMemoryCache,