[Misc] Tiny: Add test network timeouts and dynamic max-parallel for 5090/2-gpu runners (#21800)

python/sglang/test/simple_eval_common.py

@@ -516,7 +516,7 @@ def make_report_from_example_htmls(htmls: List[str]):
 def download_dataset(path, url):
     print(f"Downloading dataset {path} from {url}")
     try:
-        response = requests.get(url, stream=True)
+        response = requests.get(url, stream=True, timeout=120)
         response.raise_for_status()
 
         total_size = int(response.headers.get("content-length", 0))

python/sglang/test/simple_eval_mgsm.py

@@ -115,7 +115,7 @@ def score_mgsm(target: str, prediction: str) -> bool:
 def get_lang_examples(lang: str) -> list[dict[str, str]]:
     fpath = LANG_TO_FPATH[lang]
     examples = []
-    with urllib.request.urlopen(fpath) as f:
+    with urllib.request.urlopen(fpath, timeout=30) as f:
         for line in f.read().decode("utf-8").splitlines():
             inputs, targets = line.strip().split("\t")
             if "." in targets:

python/sglang/test/vlm_utils.py

@@ -77,7 +77,7 @@ class TestOpenAIMLLMServerBase(CustomTestCase):
         os.makedirs(cache_dir, exist_ok=True)
 
         if not os.path.exists(file_path):
-            response = requests.get(url)
+            response = requests.get(url, timeout=30)
             response.raise_for_status()
 
             with open(file_path, "wb") as f: