Unverified Commit fa3c9e06 authored by fzyzcjy's avatar fzyzcjy Committed by GitHub
Browse files

Fix popen_launch_server wait for 20 minutes when child process exits (#4777)

parent 0d658ac3
"""Common utilities for testing and benchmarking"""
import argparse
import asyncio
import copy
import logging
import os
......@@ -64,7 +63,6 @@ DEFAULT_MODEL_NAME_FOR_NIGHTLY_EVAL_QUANT_TP1 = "hugging-quants/Meta-Llama-3.1-8
DEFAULT_SMALL_MODEL_NAME_FOR_TEST_QWEN = "Qwen/Qwen2.5-1.5B-Instruct"
DEFAULT_SMALL_VLM_MODEL_NAME = "Qwen/Qwen2-VL-2B"
DEFAULT_EAGLE_TARGET_MODEL_FOR_TEST = "meta-llama/Llama-2-7b-chat-hf"
DEFAULT_EAGLE_DRAFT_MODEL_FOR_TEST = "lmsys/sglang-EAGLE-llama2-chat-7B"
......@@ -435,6 +433,11 @@ def popen_launch_server(
return process
except requests.RequestException:
pass
return_code = process.poll()
if return_code is not None:
raise Exception(f"Server unexpectedly exits ({return_code=}).")
time.sleep(10)
kill_process_tree(process.pid)
......@@ -888,7 +891,6 @@ def run_mulit_request_test(
enable_overlap=False,
chunked_prefill_size=32,
):
def workload_func(base_url, model):
def run_one(_):
prompt = """
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment