fix(e2e_test): remove dead code and fix type annotations (#16661)

sgl-model-gateway/e2e_test/conftest.py

@@ -95,7 +95,7 @@ import subprocess
 import sys
 from importlib.util import find_spec
 from pathlib import Path
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Any
 
 import pytest
 
@@ -660,8 +660,8 @@ def _get_marker_value(
     request: pytest.FixtureRequest,
     marker_name: str,
     arg_index: int = 0,
-    default: any = None,
+    default: Any = None,
-) -> any:
+) -> Any:
     """Get a value from a pytest marker.
 
     Args:
@@ -684,8 +684,8 @@ def _get_marker_value(
 def _get_marker_kwargs(
     request: pytest.FixtureRequest,
     marker_name: str,
-    defaults: dict[str, any] | None = None,
+    defaults: dict[str, Any] | None = None,
-) -> dict[str, any]:
+) -> dict[str, Any]:
     """Get keyword arguments from a pytest marker.
 
     Args:

sgl-model-gateway/e2e_test/infra/simple_eval_common.py

@@ -442,7 +442,7 @@ def make_report(eval_result: EvalResult) -> str:
     )
 
 
-def make_report_from_example_htmls(htmls: List[str]):
+def make_report_from_example_htmls(htmls: list[str]):
     """
     Create a standalone HTML report from a list of example htmls
     """

sgl-model-gateway/e2e_test/utils.py

@@ -36,21 +36,6 @@ from infra.model_specs import (  # noqa: F401; Default model paths
     _resolve_model_path,
 )
 
-# =============================================================================
-# Constants
-# =============================================================================
-
-# Server startup timeout (seconds)
-DEFAULT_TIMEOUT = 600
-DEFAULT_STARTUP_TIMEOUT = 300
-
-# Default test port range
-DEFAULT_PORT_BASE = 20000
-
-# File paths for test output
-STDOUT_FILENAME = "/tmp/sglang_test_stdout.txt"
-STDERR_FILENAME = "/tmp/sglang_test_stderr.txt"
-
 # =============================================================================
 # Tokenizer Utilities
 # =============================================================================
@@ -194,28 +179,5 @@ def is_ci_environment() -> bool:
 
 
 def get_test_timeout() -> int:
-    """Get test timeout from environment or default."""
+    """Get test timeout from environment or default (600 seconds)."""
-    return int(os.environ.get("E2E_TEST_TIMEOUT", str(DEFAULT_TIMEOUT)))
+    return int(os.environ.get("E2E_TEST_TIMEOUT", "600"))
-
-
-def skip_if_no_gpu():
-    """Skip test if no GPU is available."""
-    import pytest
-
-    try:
-        import torch
-
-        if not torch.cuda.is_available():
-            pytest.skip("No GPU available")
-    except ImportError:
-        # Try nvidia-ml-py
-        try:
-            import pynvml
-
-            pynvml.nvmlInit()
-            count = pynvml.nvmlDeviceGetCount()
-            pynvml.nvmlShutdown()
-            if count == 0:
-                pytest.skip("No GPU available")
-        except Exception:
-            pytest.skip("Cannot detect GPU (torch and pynvml not available)")