fix: Update vLLM examples routing policy

[kthui]

Overview:

Update vLLM examples routing policy for better UX and perf

Details:

• For finding prefill workers, round-robin is used.
• Turn off APC for disagg decode workers.
  • Although conditional-disagg is enabled, the max-local-prefill-length is very short, so remote prefill should be used for most requests.

Related PR: #597

Where should the reviewer start?

N/A

Related Issues: (use one of the action keywords Closes / Fixes / Resolves / Relates to)

N/A

Summary by CodeRabbit

• Chores
  • Updated routing strategy in configuration files to use "round-robin" instead of previous methods.
  • Removed prefix caching option from worker configuration.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[github-actions]

This PR is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 5 days.

[coderabbitai]

Walkthrough

The configuration files for example LLM setups were updated. The routing strategy for distributing requests among workers was changed from random or key-value-based to round-robin in both configurations. Additionally, prefix caching was disabled for VllmWorker in one configuration.

Changes

File(s)	Change Summary
examples/llm/configs/agg.yaml	Changed VllmWorker routing from "random" to "round-robin".
examples/llm/configs/disagg_router.yaml	Changed router from "kv" to "round-robin"; removed prefix caching.

Poem

Round and round, the requests now spin,
No more random, let fairness begin!
Prefix caching takes a rest,
In these configs, we route our best.
With every hop, a cycle anew—
The rabbits cheer these configs too!
🐇🔄

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

🪧 Tips

Chat

There are 3 ways to chat with CodeRabbit:

• Review comments: Directly reply to a review comment made by CodeRabbit. Example:
  • I pushed a fix in commit <commit_id>, please review it.
  • Explain this complex logic.
  • Open a follow-up GitHub issue for this discussion.
• Files and specific lines of code (under the "Files changed" tab): Tag @coderabbitai in a new review comment at the desired location with your query. Examples:
  • @coderabbitai explain this code block.
  • @coderabbitai modularize this function.
• PR comments: Tag @coderabbitai in a new PR comment to ask questions about the PR branch. For the best results, please provide a very specific query, as very limited context is provided in this mode. Examples:
  • @coderabbitai gather interesting stats about this repository and render them as a table. Additionally, render a pie chart showing the language distribution in the codebase.
  • @coderabbitai read src/utils.ts and explain its main purpose.
  • @coderabbitai read the files in the src/scheduler package and generate a class diagram using mermaid and a README in the markdown format.
  • @coderabbitai help me debug CodeRabbit configuration file.

Support

Need help? Create a ticket on our support page for assistance with any issues or questions.

Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments.

CodeRabbit Commands (Invoked using PR comments)

• @coderabbitai pause to pause the reviews on a PR.
• @coderabbitai resume to resume the paused reviews.
• @coderabbitai review to trigger an incremental review. This is useful when automatic reviews are disabled for the repository.
• @coderabbitai full review to do a full review from scratch and review all the files again.
• @coderabbitai summary to regenerate the summary of the PR.
• @coderabbitai generate docstrings to generate docstrings for this PR.
• @coderabbitai generate sequence diagram to generate a sequence diagram of the changes in this PR.
• @coderabbitai resolve resolve all the CodeRabbit review comments.
• @coderabbitai configuration to show the current CodeRabbit configuration for the repository.
• @coderabbitai help to get help.

Other keywords and placeholders

• Add @coderabbitai ignore anywhere in the PR description to prevent this PR from being reviewed.
• Add @coderabbitai summary to generate the high-level summary at a specific location in the PR description.
• Add @coderabbitai anywhere in the PR title to generate the title automatically.

CodeRabbit Configuration File (.coderabbit.yaml)

• You can programmatically configure CodeRabbit by adding a .coderabbit.yaml file to the root of your repository.
• Please see the configuration documentation for more information.
• If your editor has YAML language server enabled, you can add the path at the top of this file to enable auto-completion and validation: # yaml-language-server: $schema=https://coderabbit.ai/integrations/schema.v2.json

Documentation and Community

• Visit our Documentation for detailed information on how to use CodeRabbit.
• Join our Discord Community to get help, request features, and share feedback.
• Follow us on X/Twitter for updates and announcements.

[coderabbitai]

Actionable comments posted: 0

🔭 Outside diff range comments (1)

examples/llm/configs/agg.yaml (1)

25-40: Suggest unifying router placement across configs.

In disagg_router.yaml, router lives in Common (plus a Router section), whereas in agg.yaml it’s under Processor/VllmWorker with no Router section. Standardizing the hierarchy of routing settings across examples will reduce confusion.

🧹 Nitpick comments (1)

examples/llm/configs/disagg_router.yaml (1)

34-46: Suggest adding an explanatory comment for disagg design.

Consider annotating why conditional-disagg is enabled and prefix caching is disabled (e.g., “max-local-prefill-length is very short, so remote prefill is preferred”). This will aid future maintainers.

📜 Review details

Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between e7a49b0 and a46e17e.

📒 Files selected for processing (2)

• examples/llm/configs/agg.yaml (1 hunks)
• examples/llm/configs/disagg_router.yaml (1 hunks)

🔇 Additional comments (4)

examples/llm/configs/disagg_router.yaml (2)

17-21: Approve routing strategy update to round-robin.

Switching from "kv" to "round-robin" in the Common section ensures requests are evenly distributed across prefill workers.

---

34-38: Approve disabling prefix caching for disaggregated workers.

Removing enable-prefix-caching aligns with the intended use of remote prefill when the local prefill capacity is minimal.

examples/llm/configs/agg.yaml (2)

25-28: Verify router at the Processor level.

You added router: round-robin under Processor. Confirm that the config loader supports router definitions here and not only in Common or a dedicated Router section.

---

29-34: Approve VllmWorker routing change.

Changing the VllmWorker’s router from random to round-robin improves predictability and load balancing for aggregated decode.

[grahamking]

Examples have changed a lot since this.