Add timeout and retry params to ChatDatabricks (#165)

Signed-off-by: Sid Murching <sid.murching@databricks.com>

Author: smurching

integrations/langchain/pyproject.toml

@@ -15,14 +15,14 @@ dependencies = [
     "mlflow>=2.20.1",
     "pydantic>2.10.0",
     "unitycatalog-langchain[databricks]>=0.2.0",
+    "databricks-sdk>=0.65.0",
     "openai>=1.99.9",
 ]
 
 [project.optional-dependencies]
 dev = [
   "pytest",
   "typing_extensions",
-  "databricks-sdk>=0.34.0",
   "ruff==0.6.4",
 ]
 

integrations/langchain/src/databricks_langchain/chat_models.py

@@ -72,6 +72,8 @@ class ChatDatabricks(BaseChatModel):
                 model="databricks-claude-3-7-sonnet",
                 temperature=0,
                 max_tokens=500,
+                timeout=30.0,  # Timeout in seconds
+                max_retries=3,  # Maximum number of retries
             )
 
             # Using a WorkspaceClient instance for custom authentication
@@ -248,6 +250,10 @@ class GetPopulation(BaseModel):
     """Any extra parameters to pass to the endpoint."""
     use_responses_api: bool = False
     """Whether to use the Responses API to format inputs and outputs."""
+    timeout: Optional[float] = None
+    """Timeout in seconds for the HTTP request. If None, uses the default timeout."""
+    max_retries: Optional[int] = None
+    """Maximum number of retries for failed requests. If None, uses the default retry count."""
     client: Optional[object] = Field(default=None, exclude=True)  #: :meta private:
 
     @property
@@ -288,7 +294,14 @@ def __init__(self, **kwargs: Any):
                 )
 
         # Always use OpenAI client (supports both chat completions and responses API)
-        self.client = get_openai_client(workspace_client=self.workspace_client)
+        # Prepare kwargs for the SDK call
+        openai_kwargs = {}
+        if self.timeout is not None:
+            openai_kwargs["timeout"] = self.timeout
+        if self.max_retries is not None:
+            openai_kwargs["max_retries"] = self.max_retries
+
+        self.client = get_openai_client(workspace_client=self.workspace_client, **openai_kwargs)
 
         self.use_responses_api = kwargs.get("use_responses_api", False)
         self.extra_params = self.extra_params or {}

integrations/langchain/src/databricks_langchain/utils.py

@@ -20,23 +20,25 @@ def get_deployment_client(target_uri: str) -> Any:
         ) from e
 
 
-def get_openai_client(workspace_client: Any = None) -> Any:
+def get_openai_client(workspace_client: Any = None, **kwargs) -> Any:
     """Get an OpenAI client configured for Databricks.
 
     Args:
         workspace_client: Optional WorkspaceClient instance to use for authentication.
             If not provided, creates a default WorkspaceClient.
+        **kwargs: Additional keyword arguments to pass to get_open_ai_client(),
+            such as timeout and max_retries.
     """
     try:
         from databricks.sdk import WorkspaceClient
 
         # If workspace_client is provided, use it directly
         if workspace_client is not None:
-            return workspace_client.serving_endpoints.get_open_ai_client()
-
-        # Otherwise, create default workspace client
-        workspace_client = WorkspaceClient()
-        return workspace_client.serving_endpoints.get_open_ai_client()
+            return workspace_client.serving_endpoints.get_open_ai_client(**kwargs)
+        else:
+            # Otherwise, create default workspace client
+            workspace_client = WorkspaceClient()
+            return workspace_client.serving_endpoints.get_open_ai_client(**kwargs)
 
     except ImportError as e:
         raise ImportError(

integrations/langchain/tests/integration_tests/test_chat_models.py

@@ -715,3 +715,45 @@ def test_chat_databricks_utf8_encoding(model):
         if hasattr(chunk, "content") and chunk.content:
             full_content += chunk.content
     assert "blåbær" in full_content.lower()
+
+
+def test_chat_databricks_with_timeout_and_retries():
+    """Test that ChatDatabricks can be initialized with timeout and max_retries parameters."""
+    from unittest.mock import Mock, patch
+
+    # Mock the OpenAI client
+    mock_openai_client = Mock()
+    mock_workspace_client = Mock()
+    mock_workspace_client.serving_endpoints.get_open_ai_client.return_value = mock_openai_client
+
+    with patch("databricks.sdk.WorkspaceClient", return_value=mock_workspace_client):
+        # Create ChatDatabricks with timeout and max_retries
+        chat = ChatDatabricks(
+            model="databricks-meta-llama-3-3-70b-instruct", timeout=45.0, max_retries=3
+        )
+
+        # Verify the parameters are set correctly
+        assert chat.timeout == 45.0
+        assert chat.max_retries == 3
+
+        # Verify the client was configured with these parameters
+        assert chat.client == mock_openai_client
+
+    # Test with workspace_client parameter
+    with patch(
+        "databricks_langchain.chat_models.get_openai_client", return_value=mock_openai_client
+    ) as mock_get_client:
+        chat_with_ws = ChatDatabricks(
+            model="databricks-meta-llama-3-3-70b-instruct",
+            workspace_client=mock_workspace_client,
+            timeout=30.0,
+            max_retries=2,
+        )
+
+        # Verify get_openai_client was called with all parameters
+        mock_get_client.assert_called_once_with(
+            workspace_client=mock_workspace_client, timeout=30.0, max_retries=2
+        )
+
+        assert chat_with_ws.timeout == 30.0
+        assert chat_with_ws.max_retries == 2

integrations/langchain/tests/unit_tests/test_chat_models.py

@@ -91,6 +91,55 @@ def test_workspace_client_and_target_uri_conflict() -> None:
         )
 
 
+def test_timeout_and_max_retries_parameters() -> None:
+    """Test that timeout and max_retries parameters are properly passed to the OpenAI client."""
+    from unittest.mock import Mock, patch
+
+    mock_openai_client = Mock()
+    mock_openai_client.timeout = None
+    mock_openai_client.max_retries = None
+
+    with patch(
+        "databricks_langchain.chat_models.get_openai_client", return_value=mock_openai_client
+    ) as mock_get_client:
+        # Test with timeout and max_retries
+        llm = ChatDatabricks(model="test-model", timeout=60.0, max_retries=5)
+
+    # Verify get_openai_client was called with the correct parameters
+    mock_get_client.assert_called_once_with(workspace_client=None, timeout=60.0, max_retries=5)
+
+    # Test that client is set
+    assert llm.client == mock_openai_client
+    assert llm.timeout == 60.0
+    assert llm.max_retries == 5
+
+
+def test_timeout_and_max_retries_with_workspace_client() -> None:
+    """Test timeout and max_retries parameters work with workspace_client."""
+    from unittest.mock import Mock, patch
+
+    mock_workspace_client = Mock()
+    mock_openai_client = Mock()
+    mock_openai_client.timeout = None
+    mock_openai_client.max_retries = None
+
+    with patch(
+        "databricks_langchain.chat_models.get_openai_client", return_value=mock_openai_client
+    ) as mock_get_client:
+        llm = ChatDatabricks(
+            model="test-model", workspace_client=mock_workspace_client, timeout=30.0, max_retries=2
+        )
+
+    # Verify get_openai_client was called with all parameters
+    mock_get_client.assert_called_once_with(
+        workspace_client=mock_workspace_client, timeout=30.0, max_retries=2
+    )
+
+    assert llm.client == mock_openai_client
+    assert llm.timeout == 30.0
+    assert llm.max_retries == 2
+
+
 def test_default_workspace_client() -> None:
     """Test that default WorkspaceClient is created when none provided."""
     from unittest.mock import Mock, patch

integrations/langchain/tests/unit_tests/test_utils.py

@@ -0,0 +1,62 @@
+"""Test utilities module."""
+
+from unittest.mock import Mock, patch
+
+from databricks_langchain.utils import get_openai_client
+
+
+def test_get_openai_client_with_timeout_and_max_retries() -> None:
+    """Test that get_openai_client properly passes timeout and max_retries as kwargs to the SDK."""
+
+    mock_openai_client = Mock()
+
+    mock_workspace_client = Mock()
+    mock_workspace_client.serving_endpoints.get_open_ai_client.return_value = mock_openai_client
+
+    # Test with workspace_client, timeout, and max_retries
+    client = get_openai_client(workspace_client=mock_workspace_client, timeout=45.0, max_retries=3)
+
+    # Verify the OpenAI client was obtained with the correct kwargs
+    mock_workspace_client.serving_endpoints.get_open_ai_client.assert_called_once_with(
+        timeout=45.0, max_retries=3
+    )
+
+    # Verify the client is returned
+    assert client == mock_openai_client
+
+
+def test_get_openai_client_with_default_workspace_client() -> None:
+    """Test get_openai_client creates default WorkspaceClient when none provided."""
+
+    mock_openai_client = Mock()
+
+    mock_workspace_client = Mock()
+    mock_workspace_client.serving_endpoints.get_open_ai_client.return_value = mock_openai_client
+
+    with patch("databricks.sdk.WorkspaceClient", return_value=mock_workspace_client):
+        client = get_openai_client(timeout=30.0, max_retries=2)
+
+    # Verify default WorkspaceClient was created and kwargs were passed
+    mock_workspace_client.serving_endpoints.get_open_ai_client.assert_called_once_with(
+        timeout=30.0, max_retries=2
+    )
+
+    # Verify the client is returned
+    assert client == mock_openai_client
+
+
+def test_get_openai_client_without_timeout_and_retries() -> None:
+    """Test get_openai_client doesn't pass kwargs when not provided."""
+
+    mock_openai_client = Mock()
+
+    mock_workspace_client = Mock()
+    mock_workspace_client.serving_endpoints.get_open_ai_client.return_value = mock_openai_client
+
+    client = get_openai_client(workspace_client=mock_workspace_client)
+
+    # Verify the OpenAI client was obtained without kwargs
+    mock_workspace_client.serving_endpoints.get_open_ai_client.assert_called_once_with()
+
+    # Verify the client is returned
+    assert client == mock_openai_client