From df36845839378b3c8bff0cffd69aac5aa4bb2e1d Mon Sep 17 00:00:00 2001
From: Ryan Crabbe <rcrabbe@berkeley.edu>
Date: Thu, 26 Feb 2026 17:39:01 -0800
Subject: [PATCH] fix: remove cache eviction close that kills in-use httpx
 clients

---
 litellm/caching/llm_caching_handler.py        | 19 ----------
 .../caching/test_redis_connection_pool.py     | 36 -------------------
 2 files changed, 55 deletions(-)

diff --git a/litellm/caching/llm_caching_handler.py b/litellm/caching/llm_caching_handler.py
index 5dc16a224c..16eb824f4c 100644
--- a/litellm/caching/llm_caching_handler.py
+++ b/litellm/caching/llm_caching_handler.py
@@ -8,25 +8,6 @@ from .in_memory_cache import InMemoryCache
 
 
 class LLMClientCache(InMemoryCache):
-    def _remove_key(self, key: str) -> None:
-        """Close async clients before evicting them to prevent connection pool leaks."""
-        value = self.cache_dict.get(key)
-        super()._remove_key(key)
-        if value is not None:
-            close_fn = getattr(value, "aclose", None) or getattr(
-                value, "close", None
-            )
-            if close_fn and asyncio.iscoroutinefunction(close_fn):
-                try:
-                    asyncio.get_running_loop().create_task(close_fn())
-                except RuntimeError:
-                    pass
-            elif close_fn and callable(close_fn):
-                try:
-                    close_fn()
-                except Exception:
-                    pass
-
     def update_cache_key_with_event_loop(self, key):
         """
         Add the event loop to the cache key, to prevent event loop closed errors.
diff --git a/tests/test_litellm/caching/test_redis_connection_pool.py b/tests/test_litellm/caching/test_redis_connection_pool.py
index b8922846e8..f6e429ceff 100644
--- a/tests/test_litellm/caching/test_redis_connection_pool.py
+++ b/tests/test_litellm/caching/test_redis_connection_pool.py
@@ -4,14 +4,12 @@ Regression tests for Redis connection pool leak fixes (RC1-RC5).
 Tests are pure unit tests — no Redis server required.
 """
 
-import asyncio
 from unittest.mock import AsyncMock, MagicMock, patch
 
 import pytest
 import redis.asyncio as async_redis
 
 from litellm._redis import get_redis_async_client, get_redis_connection_pool
-from litellm.caching.llm_caching_handler import LLMClientCache
 
 
 def test_url_config_uses_passed_pool():
@@ -131,37 +129,3 @@ async def test_disconnect_idempotent():
     await cache.disconnect()  # should not raise
 
 
-@pytest.mark.asyncio
-async def test_eviction_calls_aclose():
-    """When an async client is evicted from LLMClientCache, its aclose()
-    should be scheduled via create_task."""
-    cache = LLMClientCache(max_size_in_memory=2, default_ttl=600)
-
-    client = AsyncMock()
-    client.aclose = AsyncMock()
-
-    cache.set_cache(key="client-0", value=client)
-    cache.set_cache(key="filler", value="x")
-    # Third insert triggers eviction of client-0
-    cache.set_cache(key="trigger", value="y")
-
-    # Let the scheduled task run
-    await asyncio.sleep(0.05)
-
-    assert client.aclose.await_count > 0
-
-
-@pytest.mark.asyncio
-async def test_eviction_non_closeable_safe():
-    """Evicting plain values (strings, dicts, ints) should not crash."""
-    cache = LLMClientCache(max_size_in_memory=2, default_ttl=600)
-
-    cache.set_cache(key="str-val", value="hello")
-    cache.set_cache(key="dict-val", value={"foo": "bar"})
-    # This evicts "str-val" — should not raise
-    cache.set_cache(key="int-val", value=42)
-
-    await asyncio.sleep(0.05)
-
-    # If we got here without exception, the test passes
-    assert cache.get_cache(key="int-val") == 42