[Test] L5 reliability test for wan2.2 and qwen3_omni model (#2972)

Signed-off-by: zhumingjue <zhumingjue@huawei.com>
Signed-off-by: [Your Name] <[Your Email]>
Signed-off-by: zhumingjue138 <zhumingjue@huawei.com>
This commit is contained in:
zhumingjue138 authored and GitHub committed 2026-04-27 12:12:52 +08:00
1 parent 52f17bf292
commit a150194b11
11 files changed
+2099 -93

No files matched your search

+85
View File
@@ -9,7 +9,9 @@ import socket
import subprocess
import sys
import tempfile
import threading
import time
from collections.abc import Generator
from dataclasses import dataclass
from io import BytesIO
from pathlib import Path
@@ -120,6 +122,89 @@ class OmniServerParams(NamedTuple):
stage_init_timeout: int | None = None # None: fixture supplies default (600 s)
def run_omni_server(
request: Any,
run_level: str,
model_prefix: str,
omni_fixture_lock: threading.Lock,
) -> Generator["OmniServer", Any, None]:
from tests.helpers.stage_config import modify_stage_config
with omni_fixture_lock:
params: OmniServerParams = request.param
model = model_prefix + params.model
port = params.port
stage_config_path = params.stage_config_path
if run_level in {"advanced_model", "full_model"} and stage_config_path is not None:
with open(stage_config_path, encoding="utf-8") as f:
cfg = yaml.safe_load(f) or {}
# Strip ``load_format: dummy`` (CI overlay default) so advanced_model
# tests use real weights. New schema (``stages:``) writes the field
# flat at stage level; legacy schema (``stage_args:``) nests it as
# ``engine_args.load_format``. Handle both.
new_schema_stages = cfg.get("stages")
stage_key = "stages" if new_schema_stages is not None else "stage_args"
delete_path = "load_format" if new_schema_stages is not None else "engine_args.load_format"
stage_entries = cfg.get(stage_key, [])
stage_ids = [stage["stage_id"] for stage in stage_entries if "stage_id" in stage]
stage_config_path = modify_stage_config(
stage_config_path,
deletes={stage_key: {stage_id: [delete_path] for stage_id in stage_ids}},
)
server_args = params.server_args or []
if params.use_omni and params.stage_init_timeout is not None:
server_args = [*server_args, "--stage-init-timeout", str(params.stage_init_timeout)]
else:
server_args = [*server_args, "--stage-init-timeout", "600"]
if params.init_timeout is not None:
server_args = [*server_args, "--init-timeout", str(params.init_timeout)]
else:
server_args = [*server_args, "--init-timeout", "900"]
if params.use_stage_cli:
if not params.use_omni:
raise ValueError("omni_server with use_stage_cli=True requires use_omni=True")
if stage_config_path is None:
raise ValueError("omni_server with use_stage_cli=True requires a stage_config_path")
server_args += ["--stage-configs-path", stage_config_path]
with OmniServerStageCli(
model,
stage_config_path,
server_args,
port=port,
env_dict=params.env_dict,
) as server:
print("OmniServer started successfully")
yield server
print("OmniServer stopping...")
else:
if stage_config_path is not None:
server_args += ["--stage-configs-path", stage_config_path]
with (
OmniServer(
model,
server_args,
port=port,
env_dict=params.env_dict,
use_omni=params.use_omni,
)
if port
else OmniServer(
model,
server_args,
env_dict=params.env_dict,
use_omni=params.use_omni,
)
) as server:
print("OmniServer started successfully")
yield server
print("OmniServer stopping...")
print("OmniServer stopped")
class OmniServer:
"""Omniserver for vLLM-Omni tests."""