feat: receive kvmetrics from sglang scheduler

[ishandhanani]

The bytedance-iaas team (@zixuanzhang226) has done some incredible work in wiring up receiving KV-related metrics sent from the SGLang scheduler to the Dynamo worker in #1465.

Since we reworked our components in #1721 - this PR basically ports the code they've written into our new components.

All code here should be attributed to their work in #1465. This PR will be coauthored by them as well

Summary by CodeRabbit

• New Features

  • Integrated real-time metrics updates from an external scheduler, enabling dynamic monitoring of metrics within the application.

• Improvements

  • Metrics are now updated continuously and asynchronously, providing more accurate and up-to-date information for users.

[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.

[coderabbitai]

Walkthrough

The update integrates ZeroMQ to enable real-time metric reception from an external SGL scheduler. It introduces an asynchronous loop to receive metrics via a zmq PULL socket and publish them, replacing the previous static metrics update mechanism. The setup and initialization routines are adjusted to accommodate this new dynamic metrics flow.

Changes

File(s)	Change Summary
examples/sglang/components/worker.py	Integrated zmq for real-time metrics reception and publishing; refactored setup and update logic

Sequence Diagram(s)

sequenceDiagram
    participant SGL Scheduler
    participant Worker (RequestHandler)
    participant Metrics Publisher

    SGL Scheduler->>Worker (zmq PULL socket): Send metrics
    loop Continuous
        Worker (zmq PULL socket)->>Worker: Receive metrics (async loop)
        Worker->>Metrics Publisher: Publish metrics
    end

Loading

Poem

A socket pulls, a rabbit grins,
Real-time metrics now begin!
ZeroMQ hops in the flow,
Data streams where numbers grow.
Goodbye to static, hello to new—
Metrics dance, dynamic 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 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.

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: 4

📜 Review details

Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 1630f8b and f9403db.

📒 Files selected for processing (1)

• examples/sglang/components/worker.py (4 hunks)

🧰 Additional context used

🧠 Learnings (1)

examples/sglang/components/worker.py (1)

Learnt from: PeaBrane
PR: ai-dynamo/dynamo#1392
File: launch/dynamo-run/src/subprocess/vllm_v1_inc.py:71-71
Timestamp: 2025-06-05T01:04:24.775Z
Learning: The `create_endpoint` method in `WorkerMetricsPublisher` has backward compatibility maintained through pyo3 signature annotation `#[pyo3(signature = (component, dp_rank = None))]`, making the `dp_rank` parameter optional with a default value of `None`.

🪛 GitHub Actions: Pre Merge Validation of (ai-dynamo/dynamo/refs/pull/1789/merge) by ishandhanani.

examples/sglang/components/worker.py

[error] 1-1: isort formatting check failed. The file was modified to fix import order.

---

[error] 1-1: Black formatting check failed. The file was reformatted to comply with Black style.

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)

• GitHub Check: Build and Test - vllm

🔇 Additional comments (2)

examples/sglang/components/worker.py (2)

46-48: LGTM!

The ZMQ context initialization and placeholder for the metrics socket are properly set up in the constructor.

---

88-106: Add graceful shutdown and metrics validation in _receive_and_publish_metrics_loop

To avoid unbounded loops on shutdown and guard against malformed data, update examples/sglang/components/worker.py as follows:

• Initialize and check a shutdown flag

   class YourWorker:
       def __init__(self, …):
           # … existing initialization …
+            self._shutdown = False

       async def _receive_and_publish_metrics_loop(self):
           """Receive metrics from SGL scheduler and publish them"""
-            while True:
+            while not self._shutdown:
               try:
                   kv_metrics = await self.receive_metrics_from_scheduler.recv_pyobj()

• Expose a stop method to signal loop exit

    async def stop(self):
        """Signal the metrics loop to exit and wait for pending tasks."""
        self._shutdown = True
        # optionally await the loop task here

• Validate incoming metrics before publishing

                   kv_metrics = await self.receive_metrics_from_scheduler.recv_pyobj()
+                    required = [
+                        'request_active_slots','request_total_slots',
+                        'kv_active_blocks','kv_total_blocks',
+                        'num_requests_waiting','gpu_cache_usage_perc',
+                        'gpu_prefix_cache_hit_rate'
+                    ]
+                    missing = [a for a in required if not hasattr(kv_metrics, a)]
+                    if missing:
+                        logging.error(f"Malformed metrics, missing: {missing}")
+                        continue
                   self.metrics_publisher.publish(
                       request_active_slots=kv_metrics.request_active_slots,
                       # … other fields …
                   )

• Consider catching specific errors (AttributeError, ZMQError) rather than a blanket except Exception so transient vs. persistent failures are distinguished.

These changes ensure the loop can terminate cleanly and only valid metrics reach the publisher.

⛔ Skipped due to learnings

Learnt from: PeaBrane
PR: ai-dynamo/dynamo#1285
File: lib/llm/src/kv_router/scheduler.rs:260-266
Timestamp: 2025-05-30T06:34:12.785Z
Learning: In the KV router scheduler code, PeaBrane prefers fail-fast behavior over silent failure handling. When accessing worker metrics data that could be out-of-bounds (like dp_rank indexing), explicit panics are preferred over graceful degradation with continue statements to ensure data integrity issues are caught early.