[BugFix] Fix port lookup in internal DP LB tests

[njhill]

Fix subtle race condition bug where API endpoint port is retrieved from the API server which is assumed to be first server in the list, but might not be depending on which starts first.

[gemini-code-assist]

Code Review

This pull request effectively addresses a race condition in the multi-node internal load balancer tests by pre-allocating the server list. This ensures that server information is stored in a thread-safe manner at a deterministic index. However, this change introduces a critical bug in the __exit__ methods of both MultinodeInternalLBServerManager and APIOnlyServerManager. The cleanup logic was not updated to handle the new list structure which can contain None values, leading to potential TypeErrors and zombie server processes if a server fails to start. This needs to be fixed to ensure reliable test cleanup.

[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.

🚀

[DarkLight1337]

Thanks for fixing, let's see if the test passes