[serve.llm] Remove upstreamed workarounds 1/3

[eicherseiji]

Why are these changes needed?

No longer needed since the behavior was aligned upstream:

• [Bugfix] Register reducer even if transformers_modules not available vllm-project/vllm#19510

Related issue number

Checks

• I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
• I've run scripts/format.sh to lint the changes in this PR.
• I've included any doc changes needed for https://docs.ray.io/en/master/.
  • I've added any new APIs to the API Reference. For example, if I added a
    method in Tune, I've added it in doc/source/tune/api/ under the
    corresponding .rst file.
• I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
• Testing Strategy
  • Unit tests
  • Release tests
  • This PR is not tested :(

[gemini-code-assist]

Summary of Changes

Hello @eicherseiji, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request focuses on code cleanup and maintenance by removing several temporary workarounds. These workarounds were originally introduced to mitigate issues in upstream libraries, but with recent fixes integrated into those dependencies, they are no longer necessary, simplifying our codebase.

Highlights

• Transformers Workaround Removal: I've removed the explicit init_hf_modules() call and its import from llm_engine.py. This workaround was previously needed to prevent pickle errors when using trust_remote_code=True with Hugging Face transformers models, indicating the underlying issue has been resolved upstream.
• vLLM Environment Variable Cleanup: I've eliminated the conditional deletion of the CUDA_VISIBLE_DEVICES environment variable in vllm_engine.py. This specific workaround, tracked by a TODO referencing a vLLM pull request, is now obsolete as the fix has been merged upstream.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point in your pull request via creating an issue comment (i.e. comment on the pull request page) using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in issue comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist is currently in preview and may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments to provide feedback.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request removes two workarounds that are no longer needed due to fixes in upstream libraries (transformers and vllm). The changes are correct and improve code maintainability.

[eicherseiji]

Services failing to start due to:

File "/home/ray/anaconda3/lib/python3.11/site-packages/vllm/distributed/kv_transfer/kv_connector/factory.py", line 49, in create_connector_v0\n    assert issubclass(connector_cls, KVConnectorBase)\n           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n',)('AssertionError\n',)

But seems like we're using the V0 engine since explicit env variable was removed in #54440
https://github.com/vllm-project/vllm/blob/releases/v0.9.2/vllm/engine/arg_utils.py#L1479

INFO 07-11 14:59:26 [arg_utils.py:1746] Engine in background thread is experimental on VLLM_USE_V1=1. Falling back to V0 Engine.

Service link

But nightly tests are successful, so need to determine why these changes are causing them to fail.

[github-actions]

This pull request has been automatically marked as stale because it has not had
any activity for 14 days. It will be closed in another 14 days if no further activity occurs.
Thank you for your contributions.

You can always ask for help on our discussion forum or Ray's public slack channel.

If you'd like to keep this open, just leave any comment, and the stale label will be removed.

[eicherseiji]

I suspect removing the vllm_config initialization via .remote() is responsible for the release test failures. Splitting that out in a separate PR to unblock these changes.

Next PR: #56170

[eicherseiji]

Failing release test is jailed, will be fixed with #56104

[eicherseiji]

Release test may have been a true positive. If the LMCache integration relies on setting CUDA_VISIBLE_DEVICES, we can't remove the lines setting it just yet. I'll update the comment if the test passes.