fix(warmup): reduce log noise for expected warm-up failures

Log LLMProxyError and common transport errors without exc_info on
retry attempts; keep full tracebacks for unexpected exceptions.
Add RateLimitExceededError retry classification test.

Made-with: Cursor

Author: Mateusz

src/core/services/usage_window_warmup_service.py

@@ -249,14 +249,33 @@ async def _execute_single_target(
                 )
             except asyncio.CancelledError:
                 raise
-            except Exception as exc:
+            except (LLMProxyError, ConnectionError, OSError, TimeoutError) as exc:
+                # Expected warm-up failures (rate limits, routing, transport): log the
+                # message only — full tracebacks here are noisy in production logs.
                 retryable = self._is_retryable_error(exc)
                 logger.warning(
                     "Usage window warm-up attempt %d/2 failed for '%s' target '%s': %s",
                     attempt,
                     entry.model,
                     target_label,
                     exc,
+                )
+                if not retryable or attempt >= 2:
+                    logger.error(
+                        "Usage window warm-up exhausted attempts for '%s' target '%s'",
+                        entry.model,
+                        target_label,
+                    )
+                    return
+            except Exception as exc:
+                retryable = self._is_retryable_error(exc)
+                logger.error(
+                    "Unexpected error during usage window warm-up attempt %d/2 "
+                    "for '%s' target '%s': %s",
+                    attempt,
+                    entry.model,
+                    target_label,
+                    exc,
                     exc_info=True,
                 )
                 if not retryable or attempt >= 2:

tests/unit/core/services/test_usage_window_warmup_service.py

@@ -5,7 +5,11 @@
 from unittest.mock import AsyncMock
 
 import pytest
-from src.core.common.exceptions import BackendError, RoutingError
+from src.core.common.exceptions import (
+    BackendError,
+    RateLimitExceededError,
+    RoutingError,
+)
 from src.core.domain.chat import ChatMessage
 from src.core.domain.configuration.usage_window_warmup_config import (
     UsageWindowWarmupConfig,
@@ -435,6 +439,12 @@ def test_is_retryable_error_classifies_errors_correctly(self) -> None:
             )
             is True
         )
+        assert (
+            UsageWindowWarmupService._is_retryable_error(
+                RateLimitExceededError("Rate limit reached for requests")
+            )
+            is True
+        )
 
         # LLMProxyError with retryable flag
         assert (