Merge pull request #40464 from BerriAI/litellm_internal_copy_35771

fix(azure): respect DEFAULT_MAX_RETRIES in initialize_azure_sdk_client (internal copy of #35771)

litellm/llms/azure/common_utils.py

@@ -14,6 +14,7 @@ from typing_extensions import ReadOnly, TypedDict
 import litellm
 from litellm._logging import verbose_logger
 from litellm.caching.caching import DualCache
+from litellm.constants import DEFAULT_MAX_RETRIES
 from litellm.llms.base_llm.chat.transformation import BaseLLMException
 from litellm.llms.openai.common_utils import BaseOpenAILLM
 from litellm.secret_managers.get_azure_ad_token_provider import (
@@ -582,7 +583,8 @@ class BaseAzureLLM(BaseOpenAILLM):
         if scope is None:
             scope = "https://cognitiveservices.azure.com/.default"
 
-        max_retries: Final = litellm_params.get("max_retries")
+        configured_max_retries: Final = litellm_params.get("max_retries")
+        max_retries: Final = DEFAULT_MAX_RETRIES if configured_max_retries is None else configured_max_retries
         timeout: Final = litellm_params.get("timeout")
         if not api_key and azure_ad_token_provider is None and tenant_id and client_id and client_secret:
             verbose_logger.debug("Using Azure AD Token Provider from Entra ID for Azure Auth")
@@ -642,8 +644,7 @@ class BaseAzureLLM(BaseOpenAILLM):
         else:
             azure_client_params["http_client"] = self._get_sync_http_client()
 
-        if max_retries is not None:
-            azure_client_params["max_retries"] = max_retries
+        azure_client_params["max_retries"] = max_retries
         if timeout is not None:
             azure_client_params["timeout"] = timeout
 

tests/test_litellm/llms/azure/test_azure_common_utils.py

@@ -385,6 +385,58 @@ def test_select_azure_base_url_called(setup_mocks):
     setup_mocks["select_url"].assert_called_once()
 
 
+def test_initialize_defaults_max_retries_to_litellm_default(setup_mocks):
+    result = BaseAzureLLM().initialize_azure_sdk_client(
+        litellm_params={},
+        api_key="test-api-key",
+        api_base="https://test.openai.azure.com",
+        model_name="gpt-4",
+        api_version="2023-06-01",
+        is_async=False,
+    )
+
+    assert result["max_retries"] == litellm.constants.DEFAULT_MAX_RETRIES
+
+
+@pytest.mark.parametrize(
+    "configured, expected",
+    [(0, 0), (5, 5), (None, litellm.constants.DEFAULT_MAX_RETRIES)],
+)
+def test_initialize_honors_explicit_max_retries(setup_mocks, configured, expected):
+    result = BaseAzureLLM().initialize_azure_sdk_client(
+        litellm_params={"max_retries": configured},
+        api_key="test-api-key",
+        api_base="https://test.openai.azure.com",
+        model_name="gpt-4",
+        api_version="2023-06-01",
+        is_async=False,
+    )
+
+    assert result["max_retries"] == expected
+
+
+def test_default_max_retries_env_var_reaches_azure_sdk_client():
+    import subprocess
+    import sys
+
+    code = (
+        "from litellm.llms.azure.common_utils import BaseAzureLLM\n"
+        "client = BaseAzureLLM().get_azure_openai_client("
+        "api_key='test-api-key', api_base='https://test.openai.azure.com', api_version='2024-02-01',"
+        " client=None, _is_async=True, litellm_params={}, model='gpt-4')\n"
+        "print(client.max_retries)"
+    )
+    completed = subprocess.run(
+        [sys.executable, "-c", code],
+        env={**os.environ, "DEFAULT_MAX_RETRIES": "0"},
+        capture_output=True,
+        text=True,
+        check=True,
+    )
+
+    assert completed.stdout.strip() == "0"
+
+
 @pytest.mark.parametrize(
     "call_type",
     [