Add more refactored openai test & in CI (#7284)

Author: jhinpan

python/sglang/srt/entrypoints/openai/api_server.py

@@ -36,7 +36,7 @@
 from fastapi.responses import Response
 
 from sglang.srt.disaggregation.utils import (
-    FakeBootstrapHost,
+    FAKE_BOOTSTRAP_HOST,
     register_disaggregation_server,
 )
 from sglang.srt.entrypoints.engine import Engine, _launch_subprocesses
@@ -265,7 +265,7 @@ def _wait_and_warmup(
                     "max_new_tokens": 8,
                     "ignore_eos": True,
                 },
-                "bootstrap_host": [FakeBootstrapHost] * server_args.dp_size,
+                "bootstrap_host": [FAKE_BOOTSTRAP_HOST] * server_args.dp_size,
                 # This is a hack to ensure fake transfer is enabled during prefill warmup
                 # ensure each dp rank has a unique bootstrap_room during prefill warmup
                 "bootstrap_room": [

test/srt/openai/conftest.py

@@ -12,9 +12,10 @@
 import requests
 
 from sglang.srt.utils import kill_process_tree  # reuse SGLang helper
+from sglang.test.test_utils import DEFAULT_SMALL_MODEL_NAME_FOR_TEST
 
 SERVER_MODULE = "sglang.srt.entrypoints.openai.api_server"
-DEFAULT_MODEL = "dummy-model"
+DEFAULT_MODEL = DEFAULT_SMALL_MODEL_NAME_FOR_TEST
 STARTUP_TIMEOUT = float(os.getenv("SGLANG_OPENAI_STARTUP_TIMEOUT", 120))
 
 
@@ -39,7 +40,7 @@ def _wait_until_healthy(proc: subprocess.Popen, base: str, timeout: float) -> No
 
 
 def launch_openai_server(model: str = DEFAULT_MODEL, **kw):
-    """Spawn the draft OpenAI-compatible server and wait until it’s ready."""
+    """Spawn the draft OpenAI-compatible server and wait until it's ready."""
     port = _pick_free_port()
     cmd = [
         sys.executable,
@@ -79,7 +80,7 @@ def launch_openai_server(model: str = DEFAULT_MODEL, **kw):
 
 @pytest.fixture(scope="session")
 def openai_server() -> Generator[str, None, None]:
-    """PyTest fixture that provides the server’s base URL and cleans up."""
+    """PyTest fixture that provides the server's base URL and cleans up."""
     proc, base, log_file = launch_openai_server()
     yield base
     kill_process_tree(proc.pid)