From 37420dce0b704bd4975239d1d2465d21e9005c25 Mon Sep 17 00:00:00 2001
From: Liangsheng Yin <hnyls2002@gmail.com>
Date: Tue, 24 Mar 2026 17:04:42 -0700
Subject: [PATCH] [CI] Enable failfast (`-f`) by default in `run_suite.py`
 (#21330)

---
 python/sglang/test/ci/ci_utils.py         |  8 ++++----
 test/README.md                            |  4 +++-
 test/registered/models/test_vlm_models.py | 21 +--------------------
 test/registered/vlm/test_encoder_dp.py    | 20 +-------------------
 4 files changed, 9 insertions(+), 44 deletions(-)

diff --git a/python/sglang/test/ci/ci_utils.py b/python/sglang/test/ci/ci_utils.py
index a17498e06..ca6cf5162 100644
--- a/python/sglang/test/ci/ci_utils.py
+++ b/python/sglang/test/ci/ci_utils.py
@@ -160,10 +160,12 @@ def run_unittest_files(
             )
             file_tic = time.perf_counter()
 
+            cmd = ["python3", full_path, "-f"]
+
             if capture_output:
                 # Capture output for retry decision
                 process = subprocess.Popen(
-                    ["python3", full_path],
+                    cmd,
                     stdout=subprocess.PIPE,
                     stderr=subprocess.STDOUT,
                     text=True,
@@ -175,9 +177,7 @@ def run_unittest_files(
                     output_lines.append(line)
                 process.wait()
             else:
-                process = subprocess.Popen(
-                    ["python3", full_path], stdout=None, stderr=None
-                )
+                process = subprocess.Popen(cmd, stdout=None, stderr=None)
                 process.wait()
 
             elapsed = time.perf_counter() - file_tic
diff --git a/test/README.md b/test/README.md
index 017c39216..4b10a5eae 100644
--- a/test/README.md
+++ b/test/README.md
@@ -88,7 +88,9 @@ Here is an illustration
 
 Because the system uses a custom registry and the `run_suite.py` launcher, it supports both Python's built-in [unittest](https://docs.python.org/3/library/unittest.html) and the popular [pytest](https://docs.pytest.org/en/stable/) framework.
 The basic unit is a file, and you can use either framework in your file.
-The launcher runs `python filename.py` to execute tests, so make sure your file includes the following lines. Otherwise, CI will not run it.
+The launcher runs `python filename.py -f` to execute tests with **failfast enabled by default** — the first test method failure stops the file immediately. This avoids wasting CI time on remaining tests after a failure.
+
+Make sure your file ends with **exactly** one of the following blocks. Do not add custom `argparse` or modify `sys.argv` before calling `unittest.main()` / `pytest.main()` — the CI runner appends `-f` for failfast, and custom argument parsing will break it.
 
 ```python
 # for unittest
diff --git a/test/registered/models/test_vlm_models.py b/test/registered/models/test_vlm_models.py
index e03c56e8a..7789bed9d 100644
--- a/test/registered/models/test_vlm_models.py
+++ b/test/registered/models/test_vlm_models.py
@@ -1,6 +1,4 @@
-import argparse
 import random
-import sys
 import tempfile
 import unittest
 from types import SimpleNamespace
@@ -8,7 +6,6 @@ from types import SimpleNamespace
 from sglang.srt.utils import is_hip
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.kits.mmmu_vlm_kit import (
-    DEFAULT_MEM_FRACTION_STATIC,
     MMMUMultiModelTestBase,
 )
 from sglang.test.test_utils import is_in_ci
@@ -48,20 +45,4 @@ class TestVLMModels(MMMUMultiModelTestBase):
 
 
 if __name__ == "__main__":
-    # Define and parse arguments here, before unittest.main
-    parser = argparse.ArgumentParser(description="Test VLM models")
-    parser.add_argument(
-        "--mem-fraction-static",
-        type=float,
-        help="Static memory fraction for the model",
-        default=DEFAULT_MEM_FRACTION_STATIC,
-    )
-
-    # Parse args intended for unittest
-    args = parser.parse_args()
-
-    # Store the parsed args object on the class
-    TestVLMModels.parsed_args = args
-
-    # Pass args to unittest
-    unittest.main(argv=[sys.argv[0]])
+    unittest.main()
diff --git a/test/registered/vlm/test_encoder_dp.py b/test/registered/vlm/test_encoder_dp.py
index 47adf11df..dd0f2669c 100644
--- a/test/registered/vlm/test_encoder_dp.py
+++ b/test/registered/vlm/test_encoder_dp.py
@@ -1,9 +1,7 @@
-import argparse
 import glob
 import json
 import os
 import random
-import sys
 import unittest
 from types import SimpleNamespace
 
@@ -255,20 +253,4 @@ class TestVLMEncoderDP(CustomTestCase):
 
 
 if __name__ == "__main__":
-    # Define and parse arguments here, before unittest.main
-    parser = argparse.ArgumentParser(description="Test VLM models")
-    parser.add_argument(
-        "--mem-fraction-static",
-        type=float,
-        help="Static memory fraction for the model",
-        default=DEFAULT_MEM_FRACTION_STATIC,
-    )
-
-    # Parse args intended for unittest
-    args = parser.parse_args()
-
-    # Store the parsed args object on the class
-    TestVLMEncoderDP.parsed_args = args
-
-    # Pass args to unittest
-    unittest.main(argv=[sys.argv[0]])
+    unittest.main()