[Rebase] Rebase to vllm v0.19.0 (#2475)
This commit is contained in:
1 parent
4c031580cf
commit
2804a85e58
52 files changed
+437
-142
No files matched your search
@@ -229,6 +229,24 @@ def _wait_for_port(host: str, port: int, timeout: int = 30) -> bool:
|
||||
return False
|
||||
|
||||
|
||||
def _is_mooncake_master_available() -> bool:
|
||||
"""Check if mooncake_master binary is present and can actually execute."""
|
||||
import shutil
|
||||
|
||||
binary = shutil.which("mooncake_master")
|
||||
if binary is None:
|
||||
return False
|
||||
try:
|
||||
result = subprocess.run(
|
||||
[binary, "--help"],
|
||||
capture_output=True,
|
||||
timeout=5,
|
||||
)
|
||||
return result.returncode != 127
|
||||
except (subprocess.TimeoutExpired, OSError):
|
||||
return True
|
||||
|
||||
|
||||
def _cleanup_mooncake_processes(timeout_secs: int = 5) -> None:
|
||||
"""Clean up any existing mooncake_master processes.
|
||||
|
||||
@@ -292,6 +310,8 @@ def _load_mooncake_config(host: str, rpc_port: int, http_port: int) -> str:
|
||||
@hardware_test(res={"cuda": "H100"})
|
||||
def test_bagel_text2img_mooncake_connector(run_level):
|
||||
"""Test Bagel text2img with Mooncake connector for inter-stage communication."""
|
||||
if not _is_mooncake_master_available():
|
||||
pytest.skip("mooncake_master is not available or cannot execute (missing shared libraries like libibverbs)")
|
||||
MOONCAKE_HOST = "127.0.0.1"
|
||||
MOONCAKE_RPC_PORT = _find_free_port()
|
||||
MOONCAKE_HTTP_PORT = _find_free_port()
|
||||
|
||||
@@ -63,20 +63,27 @@ def download_tokenizer():
|
||||
|
||||
|
||||
# CI stage config for H100 / MI325
|
||||
stage_configs = [get_chunk_config()]
|
||||
tokenizer_path = download_tokenizer()
|
||||
os.environ["MIMO_AUDIO_TOKENIZER_PATH"] = tokenizer_path
|
||||
# Guard module-level setup so test collection doesn't fail in environments
|
||||
# where the model cache is read-only or models aren't available.
|
||||
try:
|
||||
stage_configs = [get_chunk_config()]
|
||||
tokenizer_path = download_tokenizer()
|
||||
os.environ["MIMO_AUDIO_TOKENIZER_PATH"] = tokenizer_path
|
||||
|
||||
# Create parameter combinations for model and stage config
|
||||
test_params = [
|
||||
OmniServerParams(
|
||||
model=model,
|
||||
stage_config_path=stage_config,
|
||||
server_args=["--chat-template", CHAT_TEMPLATE_PATH],
|
||||
test_params = [
|
||||
OmniServerParams(
|
||||
model=model,
|
||||
stage_config_path=stage_config,
|
||||
server_args=["--chat-template", CHAT_TEMPLATE_PATH],
|
||||
)
|
||||
for model in models
|
||||
for stage_config in stage_configs
|
||||
]
|
||||
except Exception as exc:
|
||||
pytest.skip(
|
||||
f"MiMo-Audio online serving tests skipped: module setup failed ({type(exc).__name__}: {exc})",
|
||||
allow_module_level=True,
|
||||
)
|
||||
for model in models
|
||||
for stage_config in stage_configs
|
||||
]
|
||||
|
||||
|
||||
def get_prompt(prompt_type="text_only"):
|
||||
|
||||
Reference in new issue
Block a user