Failed to Install API Server via Pip

[iridium-soda]

I started evaluating this repository using a clean Python 3.13 container based on Debian. Unfortunately, I encountered an issue right from the start when attempting to install LightRAG using the safest approach: cloning the repository and manually installing the API server via pip. The installation process failed with the following output:

  Preparing metadata (pyproject.toml) ... error
  error: subprocess-exited-with-error
  
  × Preparing metadata (pyproject.toml) did not run successfully.
  │ exit code: 1
  ╰─> [78 lines of output]
      /tmp/pip-build-env-hnu92sc_/overlay/lib/python3.13/site-packages/setuptools/_distutils/dist.py:270: UserWarning: Unknown distribution option: 'test_suite'
        warnings.warn(msg)
      /tmp/pip-build-env-hnu92sc_/overlay/lib/python3.13/site-packages/setuptools/_distutils/dist.py:270: UserWarning: Unknown distribution option: 'tests_require'
        warnings.warn(msg)
      ...
      AttributeError: 'dict' object has no attribute '__NUMPY_SETUP__' and no __dict__ for setting new attributes
      [end of output]
  
  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

This issue prevents me from proceeding with my evaluation of LightRAG, which is intended to serve as a critical component of my personal project.

Additionally, I found that directly using the package from PyPI is not a viable option. Neither the standard package nor the API server includes the necessary dependencies, which also blocks their usability.

Could you provide insights into the cause of this issue and suggest a solution to resolve it? Any guidance on proper installation or alternative steps to use LightRAG would be greatly appreciated.

[iridium-soda]

The issue appears to be resolved by downgrading Python to version 3.10. However, having a comprehensive installation guide and detailed documentation would be highly appreciated.

[ParisNeo]

I guess the project wasn't built correctly for 3.13 and you do not have the right build tools.
I did not test it on anything higher than 3.11. If I have some time I'll try to test that with 3.13

Maybe another alternative is to make it a uv project. This will simplify everything nd streamline the installation as uv makes things easy.

[iridium-soda]

I guess the project wasn't built correctly for 3.13 and you do not have the right build tools. I did not test it on anything higher than 3.11. If I have some time I'll try to test that with 3.13

Maybe another alternative is to make it a uv project. This will simplify everything nd streamline the installation as uv makes things easy.

To enhance usability, it would be great if a production-ready Docker image could be provided. This would offer a more convenient and streamlined way to deploy the service, especially considering that an API service has already been designed for interaction.

[ParisNeo]

That was my next idea. Once they accept the new API upgrade, i'll start building a docker image.
It's a wip.

[iridium-soda]

That was my next idea. Once they accept the new API upgrade, i'll start building a docker image. It's a wip.

I look forward to utilizing the Docker image in my project and exploring its functionality.

[ParisNeo]

I am already using it with lollms and it really rocks. You set it up, then add it as a new datalake

I'm gonna add an automatic installation and adding of datalakes in the lollms-webui interface. Lollms has already all the needed interfaces to do that but up till now it uses only classic RAG. With lightrag integration, you'll be able to define and share multiple datalakes and use them (even combine them) when discussion with your AI or when the AI is doing some agentic development.

[Strophe27]

That was my next idea. Once they accept the new API upgrade, i'll start building a docker image. It's a wip.

That sounds great ! looking forward to seeing the progress whenever it’s ready. In the meantime, I’m a bit stuck with deploying LightRag in Docker, so I’ll be eagerly awaiting any updates on this. Keep up the amazing work!

[ParisNeo]

Hi there, my pull request has been accepted, so I'll start the docker development right now!
I hope I get a fully working docker image tonight.

[ParisNeo]

OK, it is done. I hope they accept the pull request made some documentation. if you have questions don't hesitate to ask.

[Strophe27]

Merci beaucoup, incroyable tout ton taf et talents (lollms…)

[edward-yuen]

Hi, I tried to run the docker container but was getting an error that the module 'past' was not found. That module is part of 'future' and I added it to the requirements.txt files which fixed the docker build.

[ParisNeo]

Install future to get past module.
Yes that's the solution