tests(llm): add explicit retry behavior tests for empty choices

openhands-sdk/openhands/sdk/llm/llm.py

@@ -528,7 +528,9 @@ def _one_attempt(**retry_kwargs) -> ModelResponse:
             # 6) telemetry
             self._telemetry.on_response(resp, raw_resp=raw_resp)
 
-            # Ensure at least one choice
+            # Ensure at least one choice.
+            # Gemini sometimes returns empty choices; we raise LLMNoResponseError here
+            # inside the retry boundary so it is retried.
             if not resp.get("choices") or len(resp["choices"]) < 1:
                 raise LLMNoResponseError(
                     "Response choices is less than 1. Response: " + str(resp)

tests/sdk/llm/test_llm_no_response_retry.py

@@ -0,0 +1,107 @@
+from unittest.mock import patch
+
+import pytest
+from litellm.types.utils import Choices, Message as LiteLLMMessage, ModelResponse, Usage
+from pydantic import SecretStr
+
+from openhands.sdk.llm import LLM, LLMResponse, Message, TextContent
+from openhands.sdk.llm.exceptions import LLMNoResponseError
+
+
+def create_mock_response(
+    content: str = "ok", response_id: str = "r-1"
+) -> ModelResponse:
+    return ModelResponse(
+        id=response_id,
+        choices=[
+            Choices(
+                finish_reason="stop",
+                index=0,
+                message=LiteLLMMessage(content=content, role="assistant"),
+            )
+        ],
+        created=1,
+        model="gpt-4o",
+        object="chat.completion",
+        system_fingerprint="t",
+        usage=Usage(prompt_tokens=1, completion_tokens=1, total_tokens=2),
+    )
+
+
+def create_empty_choices_response(response_id: str = "empty-1") -> ModelResponse:
+    return ModelResponse(
+        id=response_id,
+        choices=[],  # triggers LLMNoResponseError inside retry boundary
+        created=1,
+        model="gpt-4o",
+        object="chat.completion",
+        usage=Usage(prompt_tokens=1, completion_tokens=0, total_tokens=1),
+    )
+
+
+@pytest.fixture
+def base_llm() -> LLM:
+    return LLM(
+        usage_id="test-llm",
+        model="gpt-4o",
+        api_key=SecretStr("test_key"),
+        num_retries=2,
+        retry_min_wait=1,
+        retry_max_wait=2,
+    )
+
+
+@patch("openhands.sdk.llm.llm.litellm_completion")
+def test_no_response_retries_then_succeeds(mock_completion, base_llm: LLM) -> None:
+    mock_completion.side_effect = [
+        create_empty_choices_response("empty-1"),
+        create_mock_response("success"),
+    ]
+
+    resp = base_llm.completion(
+        messages=[Message(role="user", content=[TextContent(text="hi")])]
+    )
+
+    assert isinstance(resp, LLMResponse)
+    assert resp.message is not None
+    assert mock_completion.call_count == 2  # initial + 1 retry
+
+
+@patch("openhands.sdk.llm.llm.litellm_completion")
+def test_no_response_exhausts_retries_bubbles_llm_no_response(
+    mock_completion, base_llm: LLM
+) -> None:
+    # Always return empty choices -> keeps raising LLMNoResponseError inside retry
+    mock_completion.side_effect = [
+        create_empty_choices_response("empty-1"),
+        create_empty_choices_response("empty-2"),
+    ]
+
+    with pytest.raises(LLMNoResponseError):
+        base_llm.completion(
+            messages=[Message(role="user", content=[TextContent(text="hi")])]
+        )
+
+    # Tenacity runs function num_retries times total
+    assert mock_completion.call_count == base_llm.num_retries
+
+
+@patch("openhands.sdk.llm.llm.litellm_completion")
+def test_no_response_retry_bumps_temperature(mock_completion, base_llm: LLM) -> None:
+    # Ensure we start at 0.0 to trigger bump to 1.0 on retry
+    assert base_llm.temperature == 0.0
+
+    mock_completion.side_effect = [
+        create_empty_choices_response("empty-1"),
+        create_mock_response("ok"),
+    ]
+
+    base_llm.completion(
+        messages=[Message(role="user", content=[TextContent(text="hi")])]
+    )
+
+    # Verify that on the second call, temperature was bumped to 1.0 by RetryMixin
+    assert mock_completion.call_count == 2
+    # Grab kwargs from the second call
+    _, second_kwargs = mock_completion.call_args_list[1]
+    assert second_kwargs.get("temperature") == 1.0