94057c3d3e
PR Test (NPU) / check-changes (push) Has been cancelled
PR Test (NPU) / pr-gate (push) Has been cancelled
PR Test (NPU) / set-image-config (push) Has been cancelled
PR Test (NPU) / stage-b-test-1-npu-a2 (0) (push) Has been cancelled
PR Test (NPU) / stage-b-test-1-npu-a2 (1) (push) Has been cancelled
PR Test (NPU) / stage-b-test-2-npu-a2 (0) (push) Has been cancelled
PR Test (NPU) / stage-b-test-2-npu-a2 (1) (push) Has been cancelled
PR Test (NPU) / stage-b-test-4-npu-a3 (push) Has been cancelled
PR Test (NPU) / stage-b-test-16-npu-a3 (push) Has been cancelled
PR Test (NPU) / multimodal-gen-test-1-npu-a3 (push) Has been cancelled
PR Test (NPU) / multimodal-gen-test-2-npu-a3 (push) Has been cancelled
PR Test (Arm64) / pr-gate (push) Has been cancelled
PR Test (Arm64) / check-changes (push) Has been cancelled
PR Test (Arm64) / build-test (push) Has been cancelled
PR Test (sgl-router) / gate (push) Has been cancelled
PR Test (sgl-router) / tier-1 — lint (push) Has been cancelled
PR Test (sgl-router) / tier-2 — build + test (push) Has been cancelled
PR Test (sgl-router) / tier-3 — docker (placeholder) (push) Has been cancelled
PR Test (sgl-router) / tier-3 — k8s integration (push) Has been cancelled
PR Test (sgl-router) / tier-3 — e2e (push) Has been cancelled
PR Test (sgl-router) / finish (push) Has been cancelled
PR Test (NPU) / single-node-poc (map[name:qwen3_6_27b_w8a8_1p_in64k_out1k_50ms runner:linux-aarch64-a3-2 test_case:test/registered/ascend/performance/qwen3_6_27b/test_npu_qwen3_6_27b_w8a8_1p_in64k_out1k_50ms.py test_type:perf]) (push) Has been cancelled
PR Test (NPU) / pr-test-npu-finish (push) Has been cancelled
PR Test (Xeon) / pr-gate (push) Has been cancelled
PR Test (Xeon) / check-changes (push) Has been cancelled
PR Test (Xeon) / build-test (, xeon-gnr, base-b-test-cpu) (push) Has been cancelled
PR Test (XPU) / check-changes (push) Has been cancelled
PR Test (XPU) / pr-gate (push) Has been cancelled
PR Test (XPU) / stage-a-test-1-gpu-xpu (push) Has been cancelled
PR Test (XPU) / wait-for-stage-a (push) Has been cancelled
PR Test (XPU) / stage-b-test-1-gpu-xpu (push) Has been cancelled
PR Test (XPU) / finish (push) Has been cancelled
CI Model Inventory / build-inventory (push) Has been cancelled
Lint / lint (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark Compilation Check (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark - Manual Policy (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark - Request Processing (push) Has been cancelled
PR Benchmark (SMG Components) / Benchmark Summary (push) Has been cancelled
PR Test (SMG) / build-wheel (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on windows (x86_64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on macos (x86_64 - auto) (push) Has been cancelled
PR Test (SMG) / python-unit-tests (push) Has been cancelled
PR Test (SMG) / unit-tests (push) Has been cancelled
PR Test (SMG) / benchmarks (push) Has been cancelled
PR Test (SMG) / chat-completions (push) Has been cancelled
PR Test (SMG) / chat-completions-4gpu (push) Has been cancelled
PR Test (SMG) / e2e (push) Has been cancelled
PR Test (SMG) / docker-build-test (push) Has been cancelled
PR Test (SMG) / k8s-integration (push) Has been cancelled
PR Test (SMG) / finish (push) Has been cancelled
PR Test (SMG) / summarize-benchmarks (push) Has been cancelled
Release SGLang Model Gateway Docker Image / publish (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on macos (aarch64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (aarch64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (x86_64 - auto) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (aarch64 - musllinux_1_1) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / build on linux (x86_64 - musllinux_1_1) (push) Has been cancelled
Release SGLang Model Gateway to PyPI / Build SDist (push) Has been cancelled
Release SGLang Model Gateway to PyPI / Upload to PyPI (push) Has been cancelled
Release SGLang Kernels / build-cu129-matrix (aarch64, 12.9, 3.10, arm-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / build-cu129-matrix (x86_64, 12.9, 3.10, x64-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / release-cu129 (push) Has been cancelled
Release SGLang Kernels / build-cu130-matrix (aarch64, 13.0, 3.10, arm-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / build-cu130-matrix (x86_64, 13.0, 3.10, x64-kernel-build-node) (push) Has been cancelled
Release SGLang Kernels / release-cu130 (push) Has been cancelled
Release SGLang Kernels / build-rocm-matrix (3.10, 700) (push) Has been cancelled
Release SGLang Kernels / build-rocm-matrix (3.10, 720) (push) Has been cancelled
Release SGLang Kernels / release-rocm700 (push) Has been cancelled
Release SGLang Kernels / release-rocm720 (push) Has been cancelled
Release SGLang Kernels / build-musa43 (43, 3.10) (push) Has been cancelled
Release SGLang Kernels / release-musa43 (push) Has been cancelled
128 lines
4.2 KiB
Python
128 lines
4.2 KiB
Python
"""Connection target for HTTP benchmark scripts.
|
|
|
|
Owns the launch-vs-connect decision in one place: a benchmark only needs a base
|
|
URL, which comes either from a server we launch or one already running.
|
|
"""
|
|
|
|
import dataclasses
|
|
import multiprocessing
|
|
import os
|
|
import time
|
|
from typing import Callable, Optional
|
|
|
|
import requests
|
|
|
|
from sglang.srt.entrypoints.http_server import launch_server
|
|
from sglang.srt.server_args import ServerArgs
|
|
from sglang.srt.utils import kill_process_tree
|
|
from sglang.srt.utils.network import resolve_base_url
|
|
|
|
DEFAULT_TIMEOUT = 600
|
|
|
|
# Field defaults of ServerArgs, used to detect when --host/--port were set
|
|
# explicitly (and would be silently ignored in connect mode).
|
|
_SERVER_ARGS_DEFAULTS = {f.name: f.default for f in dataclasses.fields(ServerArgs)}
|
|
|
|
|
|
def server_is_up(base_url: str, timeout: float = DEFAULT_TIMEOUT) -> bool:
|
|
"""Return True if a server answers /v1/models with 200 at base_url."""
|
|
try:
|
|
headers = {
|
|
"Content-Type": "application/json; charset=utf-8",
|
|
}
|
|
response = requests.get(
|
|
f"{base_url}/v1/models", headers=headers, timeout=timeout
|
|
)
|
|
return response.status_code == 200
|
|
except requests.RequestException:
|
|
return False
|
|
|
|
|
|
def _launch_server_target(launch_server_func: Callable, server_args: ServerArgs):
|
|
try:
|
|
launch_server_func(server_args)
|
|
except Exception as e:
|
|
raise e
|
|
finally:
|
|
kill_process_tree(os.getpid(), include_parent=False)
|
|
|
|
|
|
def launch_or_reuse_server(launch_server_func: Callable, server_args: ServerArgs):
|
|
base_url = resolve_base_url("", server_args.host, server_args.port)
|
|
|
|
# Reuse an already-running server instead of forking a second one onto the
|
|
# occupied port, where it would orphan, compete for the GPU, and OOM.
|
|
if server_is_up(base_url, timeout=5):
|
|
print(
|
|
f"WARNING: reusing the server already running at {base_url} "
|
|
f"(--model and server-launch args ignored). Pass --base-url to silence."
|
|
)
|
|
return None, base_url
|
|
|
|
proc = multiprocessing.Process(
|
|
target=_launch_server_target,
|
|
args=(
|
|
launch_server_func,
|
|
server_args,
|
|
),
|
|
)
|
|
proc.start()
|
|
|
|
start_time = time.time()
|
|
while time.time() - start_time < DEFAULT_TIMEOUT:
|
|
# Fail fast if the server died during startup (e.g. OOM).
|
|
if not proc.is_alive():
|
|
raise RuntimeError(
|
|
f"Server process exited during startup (exit code "
|
|
f"{proc.exitcode}); see the traceback above for the cause."
|
|
)
|
|
if server_is_up(base_url):
|
|
return proc, base_url
|
|
time.sleep(10)
|
|
|
|
# Timed out: kill the half-started server so it does not linger as an orphan.
|
|
kill_process_tree(proc.pid)
|
|
raise TimeoutError("Server failed to start within the timeout period.")
|
|
|
|
|
|
@dataclasses.dataclass
|
|
class BenchEndpoint:
|
|
"""A base URL plus the lifecycle of any server we launched to back it.
|
|
``close()`` tears down a launched server; for a connected one it is a no-op.
|
|
"""
|
|
|
|
base_url: str
|
|
_proc: Optional[multiprocessing.Process] = None
|
|
|
|
def close(self) -> None:
|
|
if self._proc is not None:
|
|
kill_process_tree(self._proc.pid)
|
|
self._proc = None
|
|
|
|
|
|
def acquire_endpoint(
|
|
server_args: ServerArgs,
|
|
base_url: str = "",
|
|
launch_server_func: Callable = launch_server,
|
|
) -> BenchEndpoint:
|
|
"""Resolve the benchmark target -- the single launch-vs-connect decision.
|
|
|
|
base_url given: connect to it (server_args is ignored). base_url empty:
|
|
launch a server from server_args. Caller must close() the result.
|
|
"""
|
|
if base_url:
|
|
ignored = [
|
|
f"--{name}"
|
|
for name in ("host", "port")
|
|
if getattr(server_args, name) != _SERVER_ARGS_DEFAULTS[name]
|
|
]
|
|
if ignored:
|
|
print(
|
|
f"WARNING: --base-url is set, so {' / '.join(ignored)} (and other "
|
|
f"launch args) are ignored; benchmarking the server at {base_url}."
|
|
)
|
|
return BenchEndpoint(base_url=base_url)
|
|
|
|
proc, url = launch_or_reuse_server(launch_server_func, server_args)
|
|
return BenchEndpoint(base_url=url, _proc=proc)
|