search

cap-ntu / ML-Model-CI

MLModelCI is a complete MLOps platform for managing, converting, profiling, and deploying MLaaS (Machine Learning-as-a-Service), bridging the gap between current ML training and serving systems.
https://mlmodelci.com
Apache License 2.0
188 stars 32 forks source link

pip installation failed #199

Open HuaizhengZhang opened 3 years ago

HuaizhengZhang commented 3 years ago

Software and Hardware Versions

Ubuntu 16.04.6 LTS (GNU/Linux 4.4.0-170-generic x86_64)

Problem description

ERROR: Could not install packages due to an OSError: [Errno 2] No such file or directory: '/home/zhz/tmp/tensorrtserver/python/tensorrtserver-1.8.0-py2.py3-none-linux_x86_64.whl'

Steps to Reproduce the Problem

Expected Behavior

Other Information

Things you tried, stack traces, related issues, suggestions on how to fix it...

YuanmingLeee commented 3 years ago

Please specify your Ubuntu version and the full error log using pip.

univerone commented 3 years ago

I came into the same problem in this colab : https://colab.research.google.com/drive/1MYhUPIzmAvob1uELCfHPwhOYEGzEmeSX#scrollTo=-MlRUUz0758o I guess it's maybe because tar archive file haven't been fully extracted when setup seek for the file tensorrtserver-1.8.0-py2.py3-none-linux_x86_64.whl?

univerone commented 3 years ago

I came into the same problem in this colab : https://colab.research.google.com/drive/1MYhUPIzmAvob1uELCfHPwhOYEGzEmeSX#scrollTo=-MlRUUz0758o I guess it's maybe because tar archive file haven't been fully extracted when setup seek for the file tensorrtserver-1.8.0-py2.py3-none-linux_x86_64.whl?

https://github.com/cap-ntu/ML-Model-CI/blob/9bf403f82aa50c58c7c12ad85a3e6fceaaa6bc57/setup.py#L70

Becasue the extraction destination is by default the current working folder, we'll fix this https://docs.python.org/3/library/tarfile.html#tarfile.TarFile.extract

YuanmingLeee commented 3 years ago

Are you installing the packages in a clean environment?

Is this a pip or a conda environment?

What is your pip version and Python version? What is your setuptools version?

Is there any other information reported by pip install .? You can get the full log by pip install . --log pip.log (so the full log is in pip.log)

HuaizhengZhang commented 3 years ago

I re-connect the server and then it works.

The issue is so wired. I will create a new account to test the installation.

BTW, I have upgraded both requests and setuptools to the latest version and updated the readme.

HuaizhengZhang commented 3 years ago

Are you installing the packages in a clean environment?

Is this a pip or a conda environment?

What is your pip version and Python version? What is your setuptools version?

Is there any other information reported by pip install .? You can get the full log by pip install . --log pip.log (so the full log is in pip.log)

users should not consider so much to install our system. this is our duty to make an extremely easy installation!

HuaizhengZhang commented 3 years ago 
Successfully built modelci pytest-env torchviz py-cpuinfo termcolor fire
Installing collected packages: pyasn1, rsa, pyasn1-modules, oauthlib, cachetools, typing-extensions, requests-oauthlib, multidict, google-auth, yarl, threadpoolctl, termcolor, tensorboard-plugin-wit, onnx, markdown, hyperframe, hpack, google-auth-oauthlib, async-timeout, absl-py, toml, tensorflow-estimator, tensorboard, scikit-learn, py, opt-einsum, onnxconverter-common, keras-preprocessing, iniconfig, h2, google-pasta, gast, fsspec, fire, astunparse, aiohttp, websocket-client, tqdm, torch, tensorflow, stringcase, starlette, skl2onnx, PyYAML, pytest, pymongo, pygments, pydantic, keras2onnx, h11, grpclib, graphviz, dill, commonmark, click, xgboost, uvicorn, torchviz, torchvision, tensorrtserver, tensorflow-serving-api, rich, pytorch-lightning, pytest-env, py-cpuinfo, opencv-python, onnxruntime, onnxmltools, mongoengine, lightgbm, Jinja2, hummingbird-ml, humanize, GPUtil, fastapi, docker, betterproto, modelci
  Attempting uninstall: scikit-learn
    Found existing installation: scikit-learn 0.22.1
    Uninstalling scikit-learn-0.22.1:
      Successfully uninstalled scikit-learn-0.22.1
  Attempting uninstall: py
    Found existing installation: py 1.8.1
    Uninstalling py-1.8.1:
      Successfully uninstalled py-1.8.1
  Attempting uninstall: fsspec
    Found existing installation: fsspec 0.6.2
    Uninstalling fsspec-0.6.2:
      Successfully uninstalled fsspec-0.6.2
  Attempting uninstall: tqdm
    Found existing installation: tqdm 4.42.1
    Uninstalling tqdm-4.42.1:
      Successfully uninstalled tqdm-4.42.1
  Attempting uninstall: PyYAML
    Found existing installation: PyYAML 5.3
ERROR: Cannot uninstall 'PyYAML'. It is a distutils installed project and thus we cannot accurately determine which files belong to it which would lead to only a partial uninstall.

still many many installation issues

HuaizhengZhang commented 3 years ago

a new conda env can install modelci successfully

so many issues.

Let us discuss these issues and improve the installation next week. I will fix them first.

We should avoid these low-class issues and offer an extremely easy-to-use mlmodelci.

YuanmingLeee commented 3 years ago

I tried with a clean pip virtual env, also works. I wonder if this is the problem when installed from some existing environment

HuaizhengZhang commented 3 years ago

I tried with a clean pip virtual env, also works. I wonder if this is the problem when installed from some existing environment

No matter how, we should avoid this. I believe our dependency is not as complex as Tensorflow or Pytorch. We must figure it out.