[Bug]: 麒麟系统+昇腾910B，从0.7.1rc1 升级 0.7.3rc1报错：安装vllm0.7.3报错

[yungongzi]

Your current environment

The output of `python collect_env.py`

OS: Kylin Linux Advanced Server V10 (Lance) (aarch64)
Python：3.11.11
NPU：910B
ROCM Version: Could not collect

VLLM_USE_MODELSCOPE=True
PYTORCH_NPU_ALLOC_CONF=max_split_size_mb:256
LD_LIBRARY_PATH=/usr/local/Ascend/nnal/atb/latest/atb/cxx_abi_1/lib:/usr/local/Ascend/nnal/atb/latest/atb/cxx_abi_1/examples:/usr/local/Ascend/nnal/atb/latest/atb/cxx_abi_1/tests/atbopstest:/usr/local/Ascend/ascend-toolkit/latest/tools/aml/lib64:/usr/local/Ascend/ascend-toolkit/latest/tools/aml/lib64/plugin:/usr/local/Ascend/ascend-toolkit/latest/lib64:/usr/local/Ascend/ascend-toolkit/latest/lib64/plugin/opskernel:/usr/local/Ascend/ascend-toolkit/latest/lib64/plugin/nnengine:/usr/local/Ascend/ascend-toolkit/latest/opp/built-in/op_impl/ai_core/tbe/op_tiling/lib/linux/aarch64:/usr/local/Ascend/driver/lib64:/usr/local/Ascend/driver/lib64/common:/usr/local/Ascend/driver/lib64/driver:
NCCL_CUMEM_ENABLE=0
TORCHINDUCTOR_COMPILE_THREADS=1

🐛 Describe the bug

vllm_ascend 0.7.1rc1版本通过源码安装方式可正常安装。
从新创建环境，升级到0.7.3rc1版本过程中，安装vllm0.7.3报错

安装方法：通过源码安装

# Install vLLM
git clone --depth 1 --branch v0.7.3 https://github.com/vllm-project/vllm
cd vllm
VLLM_TARGET_DEVICE=empty pip install . --extra-index https://download.pytorch.org/whl/cpu/

报错如下：

Processing /data/llm_projects/vllm_new
  Installing build dependencies ... done
  Getting requirements to build wheel ... error
  error: subprocess-exited-with-error
  
  × Getting requirements to build wheel did not run successfully.
  │ exit code: 1
  ╰─> [31 lines of output]
      /tmp/pip-build-env-15sxdi4l/overlay/lib/python3.11/site-packages/torch/_subclasses/functional_tensor.py:295: UserWarning: Failed to initialize NumPy: No module named 'numpy' (Triggered internally at /pytorch/torch/csrc/utils/tensor_numpy.cpp:84.)
        cpu = _conversion_method_template(device=torch.device("cpu"))
      Traceback (most recent call last):
        File "/data/miniconda3/envs/vllm_ascend_new/lib/python3.11/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 389, in <module>
          main()
        File "/data/miniconda3/envs/vllm_ascend_new/lib/python3.11/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 373, in main
          json_out["return_val"] = hook(**hook_input["kwargs"])
                                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        File "/data/miniconda3/envs/vllm_ascend_new/lib/python3.11/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 143, in get_requires_for_build_wheel
          return hook(config_settings)
                 ^^^^^^^^^^^^^^^^^^^^^
        File "/tmp/pip-build-env-15sxdi4l/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 334, in get_requires_for_build_wheel
          return self._get_build_requires(config_settings, requirements=[])
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
        File "/tmp/pip-build-env-15sxdi4l/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 304, in _get_build_requires
          self.run_setup()
        File "/tmp/pip-build-env-15sxdi4l/overlay/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_s
etup
          exec(code, locals())
        File "<string>", line 641, in <module>
        File "<string>", line 502, in get_vllm_version
        File "/tmp/pip-build-env-15sxdi4l/overlay/lib/python3.11/site-packages/setuptools_scm/_get_version_impl.py", line 16
6, in get_version
          _version_missing(config)
        File "/tmp/pip-build-env-15sxdi4l/overlay/lib/python3.11/site-packages/setuptools_scm/_get_version_impl.py", line 11
7, in _version_missing
          raise LookupError(
      LookupError: setuptools-scm was unable to detect version for /data/llm_projects/vllm_new.
      
      Make sure you're either building from a fully intact git repository or PyPI tarballs. Most other sources (such as GitH
ub's tarballs, a git checkout without the .git folder) don't contain the necessary metadata and will not work.
      
      For example, if you're using pip, instead of https://github.com/user/proj/archive/master.zip use git+https://github.co
m/user/proj.git#egg=proj
      
      Alternatively, set the version with the environment variable SETUPTOOLS_SCM_PRETEND_VERSION_FOR_${NORMALIZED_DIST_NAME
} as described in https://setuptools-scm.readthedocs.io/en/latest/config.
      [end of output]
  
  note: This error originates from a subprocess, and is likely not a problem with pip.
error: subprocess-exited-with-error
                                                                                                                             
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
                                                                                                                            
note: This error originates from a subprocess, and is likely not a problem with pip.

[yungongzi]

numpy已安装版本：1.26.4

[doit-5618]

一样的错误

[wangxiyuan]

setuptools-scm only works in a git repo, I guess the vllm code in your env is not from git clone? one way fixing it is to init git version in the folder by hand IMO

git init

[doit-5618]

root@autodl-container-3e45438682-d5658c4e:~/autodl-tmp# pip list | grep vllm
vllm 0.1.dev0+d20250317.empty
vllm_ascend 0.0.0 /root/autodl-tmp/vllm-ascend-main

[doit-5618]

setuptools-scm 只能在 git repo 中使用，我猜您环境中的 VLLM 代码不是来自 ？修复它的一种方法是手动初始化文件夹中的 git 版本 IMOgit clone

git init

new problem
root@autodl-container-3e45438682-d5658c4e:~/autodl-tmp# vllm serve deepseek-ai/DeepSeek-R1-Distill-Qwen-7B/
/root/miniconda3/lib/python3.10/site-packages/torchvision/io/image.py:13: UserWarning: Failed to load image Python extension: '/root/miniconda3/lib/python3.10/site-packages/torchvision/image.so: undefined symbol: _ZN3c1017RegisterOperatorsD1Ev'If you don't plan on using image functionality from torchvision.io, you can ignore this warning. Otherwise, there might be something wrong with your environment. Did you have libjpeg or libpng installed before building torchvision from source?
warn(
INFO 03-17 15:46:41 [init.py:30] Available plugins for group vllm.platform_plugins:
INFO 03-17 15:46:41 [init.py:32] name=ascend, value=vllm_ascend:register
INFO 03-17 15:46:41 [init.py:34] all available plugins for group vllm.platform_plugins will be loaded.
INFO 03-17 15:46:41 [init.py:36] set environment variable VLLM_PLUGINS to control which plugins to load.
INFO 03-17 15:46:41 [init.py:44] plugin ascend loaded.
INFO 03-17 15:46:41 [init.py:247] Platform plugin ascend is activated
Traceback (most recent call last):
File "/root/miniconda3/bin/vllm", line 5, in
from vllm.entrypoints.cli.main import main
File "/root/miniconda3/lib/python3.10/site-packages/vllm/entrypoints/cli/main.py", line 8, in
import vllm.entrypoints.cli.benchmark.main
File "/root/miniconda3/lib/python3.10/site-packages/vllm/entrypoints/cli/benchmark/main.py", line 4, in
import vllm.entrypoints.cli.benchmark.serve
File "/root/miniconda3/lib/python3.10/site-packages/vllm/entrypoints/cli/benchmark/serve.py", line 4, in
from vllm.benchmarks.serve import add_cli_args, main
ModuleNotFoundError: No module named 'vllm.benchmarks'
[ERROR] 2025-03-17-15:46:42 (PID:104088, Device:-1, RankID:-1) ERR99999 UNKNOWN applicaiton exception

[yungongzi]

setuptools-scm only works in a git repo, I guess the vllm code in your env is not from git clone? one way fixing it is to init git version in the folder by hand IMO

git init

由于内网原因，不能访问github。使用release的0.7.3版本源码安装。
git init后安装版本如下：
vllm 0.1.dev0+d20250317.empty

[yungongzi]

通过gitee代理了一下。

通过以下命令可正常安装：

# Install vLLM
git clone --depth 1 --branch v0.7.3 https://github.com/vllm-project/vllm
cd vllm
VLLM_TARGET_DEVICE=empty pip install . --extra-index https://download.pytorch.org/whl/cpu/

# Install vllm-project/vllm-ascend from pypi.
pip install vllm-ascend==0.7.3rc1 --extra-index https://download.pytorch.org/whl/cpu/

BUT
有个modelscope的错误

ERROR 03-18 09:23:40 config.py:102] Error retrieving file list: cannot import name '_try_login' from 'modelscope.utils.hf_util' (/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/modelscope/utils/hf_util/__init__.py), retrying 1 of 2
ERROR 03-18 09:23:42 config.py:100] Error retrieving file list: cannot import name '_try_login' from 'modelscope.utils.hf_ut
il' (/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/modelscope/utils/hf_util/__init__.py)
Traceback (most recent call last):
  File "/data/llm_projects/test/examples.py", line 22, in <module>
    llm = LLM(model="Qwen/Qwen2.5-0.5B-Instruct")
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/utils.py", line 1022, in inner
    return fn(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/entrypoints/llm.py", line 242, in __init__
    self.llm_engine = self.engine_class.from_engine_args(
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/engine/llm_engine.py", line 486, in from_engine_args
    engine_config = engine_args.create_engine_config(usage_context)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/engine/arg_utils.py", line 1127, in create_engine_co
nfig
    model_config = self.create_model_config()
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/engine/arg_utils.py", line 1047, in create_model_config
    return ModelConfig(
           ^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/config.py", line 304, in __init__
    hf_config = get_config(self.model, trust_remote_code, revision,
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/transformers_utils/config.py", line 256, in get_conf
ig
    if is_gguf or file_or_path_exists(
                  ^^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/transformers_utils/config.py", line 179, in file_or_
path_exists
    return file_exists(str(model),
           ^^^^^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/transformers_utils/config.py", line 154, in file_exists
    file_list = list_repo_files(repo_id,
                ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/transformers_utils/config.py", line 143, in list_rep
o_files
    return with_retry(lookup_files, "Error retrieving file list")
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/transformers_utils/config.py", line 97, in with_retr
y                                                                                                                           
    return func()
           ^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/transformers_utils/config.py", line 130, in lookup_f
iles
    return modelscope_list_repo_files(repo_id,
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/data/miniconda3/envs/vllm_73/lib/python3.11/site-packages/vllm/transformers_utils/utils.py", line 32, in modelscope
_list_repo_files
    from modelscope.utils.hf_util import _try_login
ImportError: cannot import name '_try_login' from 'modelscope.utils.hf_util' (/data/miniconda3/envs/vllm_73/lib/python3.11/s
ite-packages/modelscope/utils/hf_util/__init__.py)

modelscope相关issue

[Yikun]

@yungongzi Yes, you can pin modelscope < 1.23.0 like: #272:

python3 -m pip install "modelscope<1.23.0"

which was resolved by vllm-project/vllm#13807 since v0.8.x

[Yikun]

I am closing this, pls let us know if you have more questions.