Refactor and optimize mooncake CI (#11162)

Signed-off-by: Shangming Cai <csmthu@gmail.com>

python/sglang/test/test_disaggregation_utils.py

 import time
+from urllib.parse import urlparse
 
 import requests
 
 from sglang.srt.utils import kill_process_tree
 from sglang.test.test_utils import (
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+    DEFAULT_URL_FOR_TEST,
     CustomTestCase,
     popen_with_error_check,
 )
@@ -13,8 +15,17 @@ from sglang.test.test_utils import (
 class TestDisaggregationBase(CustomTestCase):
     @classmethod
     def setUpClass(cls):
+        parsed_url = urlparse(DEFAULT_URL_FOR_TEST)
+        cls.base_host = parsed_url.hostname
+        base_port = str(parsed_url.port)
+        cls.lb_port = base_port
+        cls.prefill_port = f"{int(base_port) + 100}"
+        cls.decode_port = f"{int(base_port) + 200}"
+        cls.prefill_url = f"http://{cls.base_host}:{cls.prefill_port}"
+        cls.decode_url = f"http://{cls.base_host}:{cls.decode_port}"
+        cls.lb_url = f"http://{cls.base_host}:{cls.lb_port}"
+        print(f"{cls.base_host=} {cls.lb_port=} {cls.prefill_port=} {cls.decode_port=}")
         cls.process_lb, cls.process_decode, cls.process_prefill = None, None, None
-        pass
 
     @classmethod
     def launch_lb(cls):

test/srt/run_suite.py

@@ -146,12 +146,12 @@ suites = {
     ],
     "per-commit-8-gpu": [
         TestFile("hicache/test_hicache_storage_mooncake_backend.py", 400),
-        TestFile("lora/test_lora_llama4.py", 600),
-        TestFile("test_disaggregation.py", 499),
+        TestFile("lora/test_lora_llama4.py", 400),
+        TestFile("test_disaggregation.py", 600),
         TestFile("test_disaggregation_dp_attention.py", 155),
-        TestFile("test_disaggregation_different_tp.py", 155),
-        TestFile("test_disaggregation_pp.py", 60),
-        TestFile("test_full_deepseek_v3.py", 333),
+        TestFile("test_disaggregation_different_tp.py", 600),
+        TestFile("test_disaggregation_pp.py", 140),
+        TestFile("test_full_deepseek_v3.py", 550),
     ],
     "per-commit-4-gpu-b200": [
         # TestFile("test_gpt_oss_4gpu.py", 600),

test/srt/test_disaggregation.py

@@ -3,7 +3,6 @@ import os
 import time
 import unittest
 from types import SimpleNamespace
-from urllib.parse import urlparse
 
 import requests
 
@@ -14,7 +13,6 @@ from sglang.test.test_utils import (
     DEFAULT_EAGLE_TARGET_MODEL_FOR_TEST,
     DEFAULT_MODEL_NAME_FOR_TEST,
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
-    DEFAULT_URL_FOR_TEST,
     popen_launch_pd_server,
 )
 
@@ -22,17 +20,8 @@ from sglang.test.test_utils import (
 class TestDisaggregationAccuracy(TestDisaggregationBase):
     @classmethod
     def setUpClass(cls):
+        super().setUpClass()
         cls.model = DEFAULT_MODEL_NAME_FOR_TEST
-        parsed_url = urlparse(DEFAULT_URL_FOR_TEST)
-        cls.base_host = parsed_url.hostname
-        base_port = str(parsed_url.port)
-        cls.lb_port = base_port
-        cls.prefill_port = f"{int(base_port) + 100}"
-        cls.decode_port = f"{int(base_port) + 200}"
-        cls.prefill_url = f"http://{cls.base_host}:{cls.prefill_port}"
-        cls.decode_url = f"http://{cls.base_host}:{cls.decode_port}"
-        cls.lb_url = f"http://{cls.base_host}:{cls.lb_port}"
-        print(f"{cls.base_host=} {cls.lb_port=} {cls.prefill_port=} {cls.decode_port=}")
 
         # Non blocking start servers
         cls.start_prefill()
@@ -51,9 +40,9 @@ class TestDisaggregationAccuracy(TestDisaggregationBase):
             "--disaggregation-mode",
             "prefill",
             "--tp",
-            "1",
+            "2",
             "--disaggregation-ib-device",
-            "mlx5_roce0",
+            "mlx5_roce0,mlx5_roce1",
         ]
         cls.process_prefill = popen_launch_pd_server(
             cls.model,
@@ -69,11 +58,11 @@ class TestDisaggregationAccuracy(TestDisaggregationBase):
             "--disaggregation-mode",
             "decode",
             "--tp",
-            "1",
+            "2",
             "--base-gpu-id",
-            "1",
+            "2",
             "--disaggregation-ib-device",
-            "mlx5_roce1",
+            "mlx5_roce2,mlx5_roce3",
         ]
         cls.process_decode = popen_launch_pd_server(
             cls.model,
@@ -154,20 +143,11 @@ class TestDisaggregationAccuracy(TestDisaggregationBase):
 class TestDisaggregationMooncakeFailure(TestDisaggregationBase):
     @classmethod
     def setUpClass(cls):
+        super().setUpClass()
         # set DISAGGREGATION_TEST_FAILURE_PROB to simulate failure
         os.environ["DISAGGREGATION_TEST_FAILURE_PROB"] = "0.05"
 
         cls.model = DEFAULT_MODEL_NAME_FOR_TEST
-        parsed_url = urlparse(DEFAULT_URL_FOR_TEST)
-        cls.base_host = parsed_url.hostname
-        base_port = str(parsed_url.port)
-        cls.lb_port = base_port
-        cls.prefill_port = f"{int(base_port) + 100}"
-        cls.decode_port = f"{int(base_port) + 200}"
-        cls.prefill_url = f"http://{cls.base_host}:{cls.prefill_port}"
-        cls.decode_url = f"http://{cls.base_host}:{cls.decode_port}"
-        cls.lb_url = f"http://{cls.base_host}:{cls.lb_port}"
-        print(f"{cls.base_host=} {cls.lb_port=} {cls.prefill_port=} {cls.decode_port=}")
 
         # Non blocking start servers
         cls.start_prefill()
@@ -191,9 +171,9 @@ class TestDisaggregationMooncakeFailure(TestDisaggregationBase):
             "--disaggregation-mode",
             "prefill",
             "--tp",
-            "1",
+            "2",
             "--disaggregation-ib-device",
-            "mlx5_roce0",
+            "mlx5_roce0,mlx5_roce1",
         ]
         cls.process_prefill = popen_launch_pd_server(
             cls.model,
@@ -209,11 +189,11 @@ class TestDisaggregationMooncakeFailure(TestDisaggregationBase):
             "--disaggregation-mode",
             "decode",
             "--tp",
-            "1",
+            "2",
             "--base-gpu-id",
-            "1",
+            "2",
             "--disaggregation-ib-device",
-            "mlx5_roce1",
+            "mlx5_roce2,mlx5_roce3",
         ]
         cls.process_decode = popen_launch_pd_server(
             cls.model,
@@ -254,17 +234,9 @@ class TestDisaggregationMooncakeSpec(TestDisaggregationBase):
 
     @classmethod
     def setUpClass(cls):
+        super().setUpClass()
         cls.model = DEFAULT_EAGLE_TARGET_MODEL_FOR_TEST
         cls.draft_model = DEFAULT_EAGLE_DRAFT_MODEL_FOR_TEST
-        parsed_url = urlparse(DEFAULT_URL_FOR_TEST)
-        cls.base_host = parsed_url.hostname
-        base_port = str(parsed_url.port)
-        cls.lb_port = base_port
-        cls.prefill_port = f"{int(base_port) + 100}"
-        cls.decode_port = f"{int(base_port) + 200}"
-        cls.prefill_url = f"http://{cls.base_host}:{cls.prefill_port}"
-        cls.decode_url = f"http://{cls.base_host}:{cls.decode_port}"
-        cls.lb_url = f"http://{cls.base_host}:{cls.lb_port}"
         cls.spec_args = [
             "--speculative-algorithm",
             "EAGLE",
@@ -348,18 +320,9 @@ class TestDisaggregationMooncakeSpec(TestDisaggregationBase):
 class TestDisaggregationSimulatedRetract(TestDisaggregationBase):
     @classmethod
     def setUpClass(cls):
+        super().setUpClass()
         os.environ["SGLANG_TEST_RETRACT"] = "true"
         cls.model = DEFAULT_MODEL_NAME_FOR_TEST
-        parsed_url = urlparse(DEFAULT_URL_FOR_TEST)
-        cls.base_host = parsed_url.hostname
-        base_port = str(parsed_url.port)
-        cls.lb_port = base_port
-        cls.prefill_port = f"{int(base_port) + 100}"
-        cls.decode_port = f"{int(base_port) + 200}"
-        cls.prefill_url = f"http://{cls.base_host}:{cls.prefill_port}"
-        cls.decode_url = f"http://{cls.base_host}:{cls.decode_port}"
-        cls.lb_url = f"http://{cls.base_host}:{cls.lb_port}"
-        print(f"{cls.base_host=} {cls.lb_port=} {cls.prefill_port=} {cls.decode_port=}")
 
         # Non blocking start servers
         cls.start_prefill()
@@ -383,9 +346,9 @@ class TestDisaggregationSimulatedRetract(TestDisaggregationBase):
             "--disaggregation-mode",
             "prefill",
             "--tp",
-            "1",
+            "2",
             "--disaggregation-ib-device",
-            "mlx5_roce0",
+            "mlx5_roce0,mlx5_roce1",
         ]
         cls.process_prefill = popen_launch_pd_server(
             cls.model,
@@ -401,11 +364,11 @@ class TestDisaggregationSimulatedRetract(TestDisaggregationBase):
             "--disaggregation-mode",
             "decode",
             "--tp",
-            "1",
+            "2",
             "--base-gpu-id",
-            "1",
+            "2",
             "--disaggregation-ib-device",
-            "mlx5_roce1",
+            "mlx5_roce2,mlx5_roce3",
         ]
         cls.process_decode = popen_launch_pd_server(
             cls.model,

test/srt/test_disaggregation_different_tp.py

@@ -2,14 +2,13 @@ import os
 import time
 import unittest
 from types import SimpleNamespace
-from urllib.parse import urlparse
 
 from sglang.test.few_shot_gsm8k import run_eval as run_eval_few_shot_gsm8k
 from sglang.test.test_disaggregation_utils import TestDisaggregationBase
 from sglang.test.test_utils import (
+    DEFAULT_MODEL_NAME_FOR_TEST,
     DEFAULT_MODEL_NAME_FOR_TEST_MLA,
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
-    DEFAULT_URL_FOR_TEST,
     popen_launch_pd_server,
 )
 
@@ -17,21 +16,12 @@ from sglang.test.test_utils import (
 class TestDisaggregationMooncakePrefillLargerTP(TestDisaggregationBase):
     @classmethod
     def setUpClass(cls):
+        super().setUpClass()
         # Temporarily disable JIT DeepGEMM
         cls.original_jit_deepgemm = os.environ.get("SGL_ENABLE_JIT_DEEPGEMM")
         os.environ["SGL_ENABLE_JIT_DEEPGEMM"] = "false"
 
         cls.model = DEFAULT_MODEL_NAME_FOR_TEST_MLA
-        parsed_url = urlparse(DEFAULT_URL_FOR_TEST)
-        cls.base_host = parsed_url.hostname
-        base_port = str(parsed_url.port)
-        cls.lb_port = base_port
-        cls.prefill_port = f"{int(base_port) + 100}"
-        cls.decode_port = f"{int(base_port) + 200}"
-        cls.prefill_url = f"http://{cls.base_host}:{cls.prefill_port}"
-        cls.decode_url = f"http://{cls.base_host}:{cls.decode_port}"
-        cls.lb_url = f"http://{cls.base_host}:{cls.lb_port}"
-        print(f"{cls.base_host=} {cls.lb_port=} {cls.prefill_port=} {cls.decode_port=}")
 
         # Non blocking start servers
         cls.start_prefill()
@@ -50,7 +40,7 @@ class TestDisaggregationMooncakePrefillLargerTP(TestDisaggregationBase):
             "--disaggregation-mode",
             "prefill",
             "--tp",
-            "2",
+            "4",
             "--disaggregation-ib-device",
             "mlx5_roce0,mlx5_roce1",
         ]
@@ -68,11 +58,11 @@ class TestDisaggregationMooncakePrefillLargerTP(TestDisaggregationBase):
             "--disaggregation-mode",
             "decode",
             "--tp",
-            "1",
-            "--base-gpu-id",
             "2",
+            "--base-gpu-id",
+            "4",
             "--disaggregation-ib-device",
-            "mlx5_roce2",
+            "mlx5_roce4,mlx5_roce5",
         ]
         cls.process_decode = popen_launch_pd_server(
             cls.model,
@@ -100,21 +90,12 @@ class TestDisaggregationMooncakePrefillLargerTP(TestDisaggregationBase):
 class TestDisaggregationMooncakeDecodeLargerTP(TestDisaggregationBase):
     @classmethod
     def setUpClass(cls):
+        super().setUpClass()
         # Temporarily disable JIT DeepGEMM
         cls.original_jit_deepgemm = os.environ.get("SGL_ENABLE_JIT_DEEPGEMM")
         os.environ["SGL_ENABLE_JIT_DEEPGEMM"] = "false"
 
         cls.model = DEFAULT_MODEL_NAME_FOR_TEST_MLA
-        parsed_url = urlparse(DEFAULT_URL_FOR_TEST)
-        cls.base_host = parsed_url.hostname
-        base_port = str(parsed_url.port)
-        cls.lb_port = base_port
-        cls.prefill_port = f"{int(base_port) + 100}"
-        cls.decode_port = f"{int(base_port) + 200}"
-        cls.prefill_url = f"http://{cls.base_host}:{cls.prefill_port}"
-        cls.decode_url = f"http://{cls.base_host}:{cls.decode_port}"
-        cls.lb_url = f"http://{cls.base_host}:{cls.lb_port}"
-        print(f"{cls.base_host=} {cls.lb_port=} {cls.prefill_port=} {cls.decode_port=}")
 
         # Non blocking start servers
         cls.start_prefill()
@@ -133,9 +114,83 @@ class TestDisaggregationMooncakeDecodeLargerTP(TestDisaggregationBase):
             "--disaggregation-mode",
             "prefill",
             "--tp",
-            "1",
+            "2",
+            "--disaggregation-ib-device",
+            "mlx5_roce0,mlx5_roce1",
+        ]
+        cls.process_prefill = popen_launch_pd_server(
+            cls.model,
+            cls.prefill_url,
+            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+            other_args=prefill_args,
+        )
+
+    @classmethod
+    def start_decode(cls):
+        decode_args = [
+            "--trust-remote-code",
+            "--disaggregation-mode",
+            "decode",
+            "--tp",
+            "4",
+            "--base-gpu-id",
+            "4",
+            "--disaggregation-ib-device",
+            "mlx5_roce4,mlx5_roce5",
+        ]
+        cls.process_decode = popen_launch_pd_server(
+            cls.model,
+            cls.decode_url,
+            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+            other_args=decode_args,
+        )
+
+    def test_gsm8k(self):
+        args = SimpleNamespace(
+            num_shots=5,
+            data_path=None,
+            num_questions=200,
+            max_new_tokens=512,
+            parallel=128,
+            host=f"http://{self.base_host}",
+            port=int(self.lb_port),
+        )
+        metrics = run_eval_few_shot_gsm8k(args)
+        print(f"Evaluation metrics: {metrics}")
+
+        self.assertGreater(metrics["accuracy"], 0.60)
+
+
+class TestDisaggregationMooncakeMHAPrefillLargerTP(TestDisaggregationBase):
+    @classmethod
+    def setUpClass(cls):
+        super().setUpClass()
+        # Temporarily disable JIT DeepGEMM
+        cls.original_jit_deepgemm = os.environ.get("SGL_ENABLE_JIT_DEEPGEMM")
+        os.environ["SGL_ENABLE_JIT_DEEPGEMM"] = "false"
+
+        cls.model = DEFAULT_MODEL_NAME_FOR_TEST
+
+        # Non blocking start servers
+        cls.start_prefill()
+        cls.start_decode()
+
+        # Block until both
+        cls.wait_server_ready(cls.prefill_url + "/health")
+        cls.wait_server_ready(cls.decode_url + "/health")
+
+        cls.launch_lb()
+
+    @classmethod
+    def start_prefill(cls):
+        prefill_args = [
+            "--trust-remote-code",
+            "--disaggregation-mode",
+            "prefill",
+            "--tp",
+            "4",
             "--disaggregation-ib-device",
-            "mlx5_roce0",
+            "mlx5_roce0,mlx5_roce1",
         ]
         cls.process_prefill = popen_launch_pd_server(
             cls.model,
@@ -153,9 +208,83 @@ class TestDisaggregationMooncakeDecodeLargerTP(TestDisaggregationBase):
             "--tp",
             "2",
             "--base-gpu-id",
-            "1",
+            "4",
+            "--disaggregation-ib-device",
+            "mlx5_roce4,mlx5_roce5",
+        ]
+        cls.process_decode = popen_launch_pd_server(
+            cls.model,
+            cls.decode_url,
+            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+            other_args=decode_args,
+        )
+
+    def test_gsm8k(self):
+        args = SimpleNamespace(
+            num_shots=5,
+            data_path=None,
+            num_questions=200,
+            max_new_tokens=512,
+            parallel=128,
+            host=f"http://{self.base_host}",
+            port=int(self.lb_port),
+        )
+        metrics = run_eval_few_shot_gsm8k(args)
+        print(f"Evaluation metrics: {metrics}")
+
+        self.assertGreater(metrics["accuracy"], 0.60)
+
+
+class TestDisaggregationMooncakeMHADecodeLargerTP(TestDisaggregationBase):
+    @classmethod
+    def setUpClass(cls):
+        super().setUpClass()
+        # Temporarily disable JIT DeepGEMM
+        cls.original_jit_deepgemm = os.environ.get("SGL_ENABLE_JIT_DEEPGEMM")
+        os.environ["SGL_ENABLE_JIT_DEEPGEMM"] = "false"
+
+        cls.model = DEFAULT_MODEL_NAME_FOR_TEST
+
+        # Non blocking start servers
+        cls.start_prefill()
+        cls.start_decode()
+
+        # Block until both
+        cls.wait_server_ready(cls.prefill_url + "/health")
+        cls.wait_server_ready(cls.decode_url + "/health")
+
+        cls.launch_lb()
+
+    @classmethod
+    def start_prefill(cls):
+        prefill_args = [
+            "--trust-remote-code",
+            "--disaggregation-mode",
+            "prefill",
+            "--tp",
+            "2",
+            "--disaggregation-ib-device",
+            "mlx5_roce0,mlx5_roce1",
+        ]
+        cls.process_prefill = popen_launch_pd_server(
+            cls.model,
+            cls.prefill_url,
+            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
+            other_args=prefill_args,
+        )
+
+    @classmethod
+    def start_decode(cls):
+        decode_args = [
+            "--trust-remote-code",
+            "--disaggregation-mode",
+            "decode",
+            "--tp",
+            "4",
+            "--base-gpu-id",
+            "4",
             "--disaggregation-ib-device",
-            "mlx5_roce1,mlx5_roce2",
+            "mlx5_roce4,mlx5_roce5",
         ]
         cls.process_decode = popen_launch_pd_server(
             cls.model,

test/srt/test_disaggregation_dp_attention.py

@@ -17,21 +17,12 @@ from sglang.test.test_utils import (
 class TestDisaggregationDPAttention(TestDisaggregationBase):
     @classmethod
     def setUpClass(cls):
+        super().setUpClass()
         # Temporarily disable JIT DeepGEMM
         cls.original_jit_deepgemm = os.environ.get("SGL_ENABLE_JIT_DEEPGEMM")
         os.environ["SGL_ENABLE_JIT_DEEPGEMM"] = "false"
 
         cls.model = DEFAULT_MODEL_NAME_FOR_TEST_MLA
-        parsed_url = urlparse(DEFAULT_URL_FOR_TEST)
-        cls.base_host = parsed_url.hostname
-        base_port = str(parsed_url.port)
-        cls.lb_port = base_port
-        cls.prefill_port = f"{int(base_port) + 100}"
-        cls.decode_port = f"{int(base_port) + 200}"
-        cls.prefill_url = f"http://{cls.base_host}:{cls.prefill_port}"
-        cls.decode_url = f"http://{cls.base_host}:{cls.decode_port}"
-        cls.lb_url = f"http://{cls.base_host}:{cls.lb_port}"
-        print(f"{cls.base_host=} {cls.lb_port=} {cls.prefill_port=} {cls.decode_port=}")
 
         # Non blocking start servers
         cls.start_prefill()

test/srt/test_disaggregation_pp.py

 import time
 import unittest
 from types import SimpleNamespace
-from urllib.parse import urlparse
 
 from sglang.test.few_shot_gsm8k import run_eval
 from sglang.test.test_disaggregation_utils import TestDisaggregationBase
 from sglang.test.test_utils import (
     DEFAULT_MODEL_NAME_FOR_TEST,
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
-    DEFAULT_URL_FOR_TEST,
     popen_launch_pd_server,
 )
 
@@ -16,17 +14,8 @@ from sglang.test.test_utils import (
 class TestDisaggregationPPAccuracy(TestDisaggregationBase):
     @classmethod
     def setUpClass(cls):
+        super().setUpClass()
         cls.model = DEFAULT_MODEL_NAME_FOR_TEST
-        parsed_url = urlparse(DEFAULT_URL_FOR_TEST)
-        cls.base_host = parsed_url.hostname
-        base_port = str(parsed_url.port)
-        cls.lb_port = base_port
-        cls.prefill_port = f"{int(base_port) + 100}"
-        cls.decode_port = f"{int(base_port) + 200}"
-        cls.prefill_url = f"http://{cls.base_host}:{cls.prefill_port}"
-        cls.decode_url = f"http://{cls.base_host}:{cls.decode_port}"
-        cls.lb_url = f"http://{cls.base_host}:{cls.lb_port}"
-        print(f"{cls.base_host=} {cls.lb_port=} {cls.prefill_port=} {cls.decode_port=}")
 
         # Non blocking start servers
         cls.start_prefill()
@@ -45,7 +34,7 @@ class TestDisaggregationPPAccuracy(TestDisaggregationBase):
             "--disaggregation-mode",
             "prefill",
             "--tp-size",
-            "1",
+            "2",
             "--pp-size",
             "2",
             "--disaggregation-ib-device",
@@ -66,11 +55,11 @@ class TestDisaggregationPPAccuracy(TestDisaggregationBase):
             "--disaggregation-mode",
             "decode",
             "--tp",
-            "1",
-            "--base-gpu-id",
             "2",
+            "--base-gpu-id",
+            "4",
             "--disaggregation-ib-device",
-            "mlx5_roce2",
+            "mlx5_roce4,mlx5_roce5",
         ]
         cls.process_decode = popen_launch_pd_server(
             cls.model,