[Bugfix] V1 Fix the cursor leakage issue during request scheduling.

[CLFutureX]

Background: When iterating through running requests, if preemption occurs and the preempted request is before the current cursor req_index, subsequent requests will be missed during scheduling.

Solution: When the preempted request is determined to be before the current cursor req_index, adjust the cursor to move forward to avoid missing requests.

[gemini-code-assist]

Code Review

The pull request aims to fix a bug in the scheduler where requests could be missed during preemption. The proposed logic correctly identifies the condition for adjusting the loop cursor. However, the implementation contains a critical bug that will cause a TypeError at runtime when trying to find the index of the preempted request. I've provided a suggestion to fix this issue.

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

[njhill]

Thanks @CLFutureX, good catch!

[njhill]

cc @WoosukKwon to confirm

[njhill]

@CLFutureX would you mind rebasing on latest main? I think the test failures are transient things that should now be fixed.

[WoosukKwon]

Thanks for the PR! Can we add a test about this?

[CLFutureX]

@CLFutureX I think the rebase wasn't done quite right - the PR shouldn't have all of these commits

yes, I'm sorry. I accidentally rebased the code onto the wrong branch yesterday, but I've since reverted it and adjusted the branch properly.