From bea22df6c27b30d060e6f7e239e94396c9f38ad7 Mon Sep 17 00:00:00 2001
From: yassin <yassin@berri.ai>
Date: Mon, 14 Sep 2026 19:59:10 +0000
Subject: [PATCH] fix(caching): keep generic add_cache failures at ERROR unless
 the backend is Redis

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---
 litellm/caching/caching.py                 | 13 ++++++++---
 tests/test_litellm/caching/test_caching.py | 26 ++++++++++++++++++++++
 2 files changed, 36 insertions(+), 3 deletions(-)

diff --git a/litellm/caching/caching.py b/litellm/caching/caching.py
index d6dd2a073af..be82f5def1f 100644
--- a/litellm/caching/caching.py
+++ b/litellm/caching/caching.py
@@ -679,7 +679,14 @@ class Cache:
             cache_key, cached_data, kwargs = self._add_cache_logic(result=result, **kwargs)
             self.cache.set_cache(cache_key, cached_data, **kwargs)
         except Exception as e:
-            log_redis_failure(verbose_logger, logging.ERROR, "LiteLLM Cache: exception in add_cache", e)
+            self._log_add_cache_failure(e)
+
+    def _log_add_cache_failure(self, exc: Exception) -> None:
+        message: Final = "LiteLLM Cache: exception in add_cache"
+        if isinstance(self.cache, RedisCache):
+            log_redis_failure(verbose_logger, logging.ERROR, message, exc)
+            return
+        verbose_logger.error("%s: %s", message, exc)
 
     async def async_add_cache(self, result, dynamic_cache_object: BaseCache | None = None, **kwargs):
         """
@@ -698,7 +705,7 @@ class Cache:
                 else:
                     await self.cache.async_set_cache(cache_key, cached_data, **kwargs)
         except Exception as e:
-            log_redis_failure(verbose_logger, logging.ERROR, "LiteLLM Cache: exception in add_cache", e)
+            self._log_add_cache_failure(e)
 
     def _convert_to_cached_embedding(
         self,
@@ -877,7 +884,7 @@ class Cache:
             else:
                 await self.cache.async_set_cache_pipeline(cache_list=cache_list, **kwargs)
         except Exception as e:
-            log_redis_failure(verbose_logger, logging.ERROR, "LiteLLM Cache: exception in add_cache", e)
+            self._log_add_cache_failure(e)
 
     def should_use_cache(self, **kwargs):
         """
diff --git a/tests/test_litellm/caching/test_caching.py b/tests/test_litellm/caching/test_caching.py
index 4d0ec0fb677..c7ec8abc31e 100644
--- a/tests/test_litellm/caching/test_caching.py
+++ b/tests/test_litellm/caching/test_caching.py
@@ -1,9 +1,12 @@
 import logging
 import re
+from unittest.mock import MagicMock
 
 import pytest
 
+import litellm.caching.redis_cache as redis_cache_module
 from litellm.caching.caching import Cache
+from litellm.caching.redis_cache import RedisCache, _RedisTimeoutLogThrottle
 from litellm.types.caching import LiteLLMCacheType, SemanticCacheScope
 from litellm.types.utils import Embedding, EmbeddingResponse, Usage
 
@@ -53,6 +56,29 @@ def test_cache_key_debug_log_does_not_include_prompt_material(caplog):
     assert any(cache_key in message for message in created_cache_key_logs)
 
 
+@pytest.mark.parametrize(
+    ("backend", "expected_level"),
+    [
+        pytest.param(MagicMock(spec=RedisCache), logging.DEBUG, id="redis_backend_is_throttled"),
+        pytest.param(MagicMock(), logging.ERROR, id="other_backend_logs_every_timeout"),
+    ],
+)
+def test_add_cache_timeout_only_joins_redis_throttle_for_redis_backends(backend, expected_level, caplog, monkeypatch):
+    throttle = _RedisTimeoutLogThrottle(interval=5.0, clock=MagicMock(return_value=1_000.0))
+    assert throttle.admit() == 0
+    monkeypatch.setattr(redis_cache_module, "_redis_timeout_log_throttle", throttle)
+
+    cache = Cache(type=LiteLLMCacheType.LOCAL)
+    backend.set_cache.side_effect = TimeoutError("lit7520 backend timed out")
+    cache.cache = backend
+
+    with caplog.at_level(logging.DEBUG, logger="LiteLLM"):
+        cache.add_cache("result", model="gpt-4.1-mini", messages=[{"role": "user", "content": "hi"}])
+
+    records = [r for r in caplog.records if "lit7520 backend timed out" in r.getMessage()]
+    assert [r.levelno for r in records] == [expected_level]
+
+
 def _embedding_response(prompt_tokens, num_items):
     return EmbeddingResponse(
         model="amazon.titan-embed-image-v1",