EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[CI] Refactor disaggregation tests (#10068)

Browse Source 
Signed-off-by: Shangming Cai <csmthu@gmail.com>
This commit is contained in:
Shangming Cai
2025-09-06 22:14:46 +08:00
committed by GitHub
parent 5f1eb20484
commit 00974e4f6e
5 changed files with 100 additions and 353 deletions
Download Patch File Download Diff File Expand all files Collapse all files

117
test/srt/test_disaggregation_different_tp.py
View File

@@ -1,25 +1,20 @@
import os
import subprocess
import time
import unittest
from types import SimpleNamespace
from urllib.parse import urlparse
import requests
from sglang.srt.utils import kill_process_tree
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_MLA,
DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
DEFAULT_URL_FOR_TEST,
CustomTestCase,
popen_launch_pd_server,
popen_with_error_check,
)
class TestDisaggregationMooncakePrefillLargerTP(CustomTestCase):
class TestDisaggregationMooncakePrefillLargerTP(TestDisaggregationBase):
@classmethod
def setUpClass(cls):
# Temporarily disable JIT DeepGEMM
@@ -46,25 +41,7 @@ class TestDisaggregationMooncakePrefillLargerTP(CustomTestCase):
cls.wait_server_ready(cls.prefill_url + "/health")
cls.wait_server_ready(cls.decode_url + "/health")
lb_command = [
"python3",
"-m",
"sglang_router.launch_router",
"--pd-disaggregation",
"--mini-lb", # FIXME: remove this
"--prefill",
cls.prefill_url,
"--decode",
cls.decode_url,
"--host",
cls.base_host,
"--port",
cls.lb_port,
]
print("Starting load balancer:", " ".join(lb_command))
cls.process_lb = popen_with_error_check(lb_command)
cls.wait_server_ready(cls.lb_url + "/health")
cls.launch_lb()
@classmethod
def start_prefill(cls):
@@ -104,39 +81,6 @@ class TestDisaggregationMooncakePrefillLargerTP(CustomTestCase):
other_args=decode_args,
)
@classmethod
def wait_server_ready(cls, url, timeout=60):
start_time = time.perf_counter()
while True:
try:
response = requests.get(url)
if response.status_code == 200:
print(f"Server {url} is ready")
return
except Exception:
pass
if time.perf_counter() - start_time > timeout:
raise RuntimeError(f"Server {url} failed to start in {timeout}s")
time.sleep(1)
@classmethod
def tearDownClass(cls):
# Restore JIT DeepGEMM environment variable
if cls.original_jit_deepgemm is not None:
os.environ["SGL_ENABLE_JIT_DEEPGEMM"] = cls.original_jit_deepgemm
else:
os.environ.pop("SGL_ENABLE_JIT_DEEPGEMM", None)
for process in [cls.process_lb, cls.process_decode, cls.process_prefill]:
if process:
try:
kill_process_tree(process.pid)
except Exception as e:
print(f"Error killing process {process.pid}: {e}")
# wait for 5 seconds
time.sleep(5)
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,
@@ -153,7 +97,7 @@ class TestDisaggregationMooncakePrefillLargerTP(CustomTestCase):
self.assertGreater(metrics["accuracy"], 0.60)
class TestDisaggregationMooncakeDecodeLargerTP(CustomTestCase):
class TestDisaggregationMooncakeDecodeLargerTP(TestDisaggregationBase):
@classmethod
def setUpClass(cls):
# Temporarily disable JIT DeepGEMM
@@ -180,25 +124,7 @@ class TestDisaggregationMooncakeDecodeLargerTP(CustomTestCase):
cls.wait_server_ready(cls.prefill_url + "/health")
cls.wait_server_ready(cls.decode_url + "/health")
lb_command = [
"python3",
"-m",
"sglang_router.launch_router",
"--pd-disaggregation",
"--mini-lb", # FIXME: remove this
"--prefill",
cls.prefill_url,
"--decode",
cls.decode_url,
"--host",
cls.base_host,
"--port",
cls.lb_port,
]
print("Starting load balancer:", " ".join(lb_command))
cls.process_lb = popen_with_error_check(lb_command)
cls.wait_server_ready(cls.lb_url + "/health")
cls.launch_lb()
@classmethod
def start_prefill(cls):
@@ -238,39 +164,6 @@ class TestDisaggregationMooncakeDecodeLargerTP(CustomTestCase):
other_args=decode_args,
)
@classmethod
def wait_server_ready(cls, url, timeout=60):
start_time = time.perf_counter()
while True:
try:
response = requests.get(url)
if response.status_code == 200:
print(f"Server {url} is ready")
return
except Exception:
pass
if time.perf_counter() - start_time > timeout:
raise RuntimeError(f"Server {url} failed to start in {timeout}s")
time.sleep(1)
@classmethod
def tearDownClass(cls):
# Restore JIT DeepGEMM environment variable
if cls.original_jit_deepgemm is not None:
os.environ["SGL_ENABLE_JIT_DEEPGEMM"] = cls.original_jit_deepgemm
else:
os.environ.pop("SGL_ENABLE_JIT_DEEPGEMM", None)
for process in [cls.process_lb, cls.process_decode, cls.process_prefill]:
if process:
try:
kill_process_tree(process.pid)
except Exception as e:
print(f"Error killing process {process.pid}: {e}")
# wait for 5 seconds
time.sleep(5)
def test_gsm8k(self):
args = SimpleNamespace(
num_shots=5,