Fix CI && python3.8 compatible (#920)

test/srt/test_chunked_prefill.py

@@ -11,7 +11,7 @@ class TestAccuracy(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
         cls.model = MODEL_NAME_FOR_TEST
-        cls.base_url = f"http://localhost:30000"
+        cls.base_url = f"http://localhost:8157"
         cls.process = popen_launch_server(
             cls.model,
             cls.base_url,

test/srt/test_eval_accuracy.py

@@ -11,7 +11,7 @@ class TestAccuracy(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
         cls.model = MODEL_NAME_FOR_TEST
-        cls.base_url = f"http://localhost:30000"
+        cls.base_url = f"http://localhost:8157"
         cls.process = popen_launch_server(cls.model, cls.base_url, timeout=300)
 
     @classmethod

test/srt/test_openai_server.py

@@ -12,7 +12,7 @@ class TestOpenAIServer(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
         cls.model = MODEL_NAME_FOR_TEST
-        cls.base_url = f"http://localhost:30000"
+        cls.base_url = f"http://localhost:8157"
         cls.api_key = "sk-123456"
         cls.process = popen_launch_server(
             cls.model, cls.base_url, timeout=300, api_key=cls.api_key

test/srt/test_srt_endpoint.py

@@ -12,11 +12,9 @@ class TestSRTEndpoint(unittest.TestCase):
 
     @classmethod
     def setUpClass(cls):
-        port = 30000
-
         cls.model = MODEL_NAME_FOR_TEST
-        cls.base_url = f"http://localhost:{port}"
-        cls.process = popen_launch_server(cls.model, port, timeout=300)
+        cls.base_url = f"http://localhost:{8157}"
+        cls.process = popen_launch_server(cls.model, cls.base_url, timeout=300)
 
     @classmethod
     def tearDownClass(cls):

test/srt/test_torch_compile.py

@@ -11,7 +11,7 @@ class TestAccuracy(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
         cls.model = MODEL_NAME_FOR_TEST
-        cls.base_url = f"http://localhost:30000"
+        cls.base_url = f"http://localhost:8157"
         cls.process = popen_launch_server(
             cls.model, cls.base_url, timeout=300, other_args=["--enable-torch-compile"]
         )