biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Misc] Deprecation Warning when setting --engine-use-ray (#7424)

Browse Source 
Signed-off-by: Wallas Santos <wallashss@ibm.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: Nick Hill <nickhill@us.ibm.com>
Co-authored-by: youkaichao <youkaichao@126.com>
This commit is contained in:
Wallas Henrique
2024-08-14 13:44:27 -03:00
committed by GitHub
parent 67d115db08
commit 70b746efcf
7 changed files with 56 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
tests/async_engine/test_api_server.py
View File

@@ -1,3 +1,4 @@
import os
import subprocess
import sys
import time
@@ -35,11 +36,17 @@ def api_server(tokenizer_pool_size: int, engine_use_ray: bool,
"127.0.0.1", "--tokenizer-pool-size",
str(tokenizer_pool_size)
]
# Copy the environment variables and append `VLLM_ALLOW_ENGINE_USE_RAY=1`
# to prevent `--engine-use-ray` raises an exception due to it deprecation
env_vars = os.environ.copy()
env_vars["VLLM_ALLOW_ENGINE_USE_RAY"] = "1"
if engine_use_ray:
commands.append("--engine-use-ray")
if worker_use_ray:
commands.append("--worker-use-ray")
uvicorn_process = subprocess.Popen(commands)
uvicorn_process = subprocess.Popen(commands, env=env_vars)
yield
uvicorn_process.terminate()

6
tests/async_engine/test_async_llm_engine.py
View File

@@ -1,4 +1,5 @@
import asyncio
import os
from dataclasses import dataclass
import pytest
@@ -106,11 +107,16 @@ async def test_new_requests_event():
assert engine.engine.add_request_calls == 3
assert engine.engine.step_calls == old_step_calls + 1
# Allow deprecated engine_use_ray to not raise exception
os.environ["VLLM_ALLOW_ENGINE_USE_RAY"] = "1"
engine = MockAsyncLLMEngine(worker_use_ray=True, engine_use_ray=True)
assert engine.get_model_config() is not None
assert engine.get_tokenizer() is not None
assert engine.get_decoding_config() is not None
os.environ.pop("VLLM_ALLOW_ENGINE_USE_RAY")
def test_asyncio_run():
wait_for_gpu_memory_to_clear(

6
tests/async_engine/test_openapi_server_ray.py
View File

@@ -23,7 +23,11 @@ def server():
str(chatml_jinja_path),
]
with RemoteOpenAIServer(MODEL_NAME, args) as remote_server:
# Allow `--engine-use-ray`, otherwise the launch of the server throw
# an error due to try to use a deprecated feature
env_dict = {"VLLM_ALLOW_ENGINE_USE_RAY": "1"}
with RemoteOpenAIServer(MODEL_NAME, args,
env_dict=env_dict) as remote_server:
yield remote_server