Skip to content

Conversation

@dmitry-tokarev-nv
Copy link
Contributor

@dmitry-tokarev-nv dmitry-tokarev-nv commented Aug 22, 2025

Overview:

Cherry-pick for #2611

Details:

Where should the reviewer start?

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

  • closes GitHub issue: #xxx

Summary by CodeRabbit

  • New Features

    • Switched default example/model configs to Qwen/Qwen3-0.6B across sglang and TensorRT-LLM deployments and launch scripts.
    • Added Prometheus to vLLM runtime images.
  • Bug Fixes

    • Improved sglang robustness: safer handling when token IDs are missing and automatic tokenizer init behavior.
    • Refined GPU resource configuration for vLLM decode workers.
  • Documentation

    • Updated setup guides, health checks, and examples; simplified multimodal instructions; refreshed HiCache guidance; version notes updated.
  • Chores

    • Upgraded base images and dependencies (TensorRT-LLM rc6, PyTorch stack, UCX pinning).
  • Tests

    • Updated model references, increased timeouts, and narrowed markers.

@copy-pr-bot
Copy link

copy-pr-bot bot commented Aug 22, 2025

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 github-actions bot added the fix label Aug 22, 2025
@dmitry-tokarev-nv dmitry-tokarev-nv changed the base branch from main to release/0.4.1 August 22, 2025 02:31
@dmitry-tokarev-nv dmitry-tokarev-nv merged commit 2429b48 into release/0.4.1 Aug 22, 2025
3 of 4 checks passed
@dmitry-tokarev-nv dmitry-tokarev-nv deleted the dtokarev-0.4.1-disable-kvbm-tests branch August 22, 2025 02:31
@dmitry-tokarev-nv dmitry-tokarev-nv changed the title fix: 0.4.1 disable kvbm tests fix: 0.4.1 disable kvbm tests (CP #2611) Aug 22, 2025
@coderabbitai
Copy link
Contributor

coderabbitai bot commented Aug 22, 2025

Caution

Review failed

The pull request is closed.

Walkthrough

Bulk updates switch default model references to Qwen/Qwen3-0.6B, bump TensorRT-LLM pins to 1.0.0rc6 with container/version alignments, adjust UCX refs to v1.19.0, remove the local async-openai-macros crate in favor of a published dependency, and add SGLang runtime behavior tweaks (frontend tokenization default, decode error guard). Tests and docs updated accordingly.

Changes

Cohort / File(s) Summary of changes
Rust macros removal & workspace update
Cargo.toml, lib/async-openai-macros/Cargo.toml, lib/async-openai-macros/src/lib.rs, lib/async-openai/Cargo.toml
Removed local async-openai-macros crate from workspace and repo; lib/async-openai now depends on published async-openai-macros = "0.1.0".
SGLang model/config/docs switch
components/backends/sglang/README.md, components/backends/sglang/deploy/*.yaml, components/backends/sglang/launch/*.sh, components/backends/sglang/docs/*
Replaced model identifiers from deepseek-ai/DeepSeek-R1-Distill-Llama-8B to Qwen/Qwen3-0.6B; added --skip-tokenizer-init where noted; unified some invocation entrypoints; HiCache example switched from size to ratio.
SGLang runtime behavior
components/backends/sglang/src/dynamo/sglang/args.py, .../request_handlers/decode_handler.py, components/backends/sglang/slurm_jobs/scripts/worker_setup.py, components/backends/sglang/slurm_jobs/scripts/h100.sh
Default enforce skip_tokenizer_init=True with warning; guarded missing output_ids with descriptive error; changed frontend startup command; adjusted --cuda-graph-bs 256→128 in H100 script.
TRT-LLM model/config/docs switch
components/backends/trtllm/deploy/*.yaml, components/backends/trtllm/launch/*.sh, components/backends/trtllm/README.md, components/backends/trtllm/deploy/README.md, components/backends/trtllm/gemma3_sliding_window_attention.md, components/backends/trtllm/gpt-oss.md
Switched model to Qwen/Qwen3-0.6B in deploy/launch; cleaned MTP sections; added readiness health-check guidance; updated notes.
Container/tooling pins (UCX/TRT-LLM/PyTorch)
container/Dockerfile*, container/build.sh, pyproject.toml, README.md, docs/support_matrix.md
Bumped UCX ref to v1.19.0; updated TRT-LLM base tag to 25.06, wheel to 1.0.0rc6, related pins (Torch, FlashAttention, NetworkX, CUDA runtime); introduced/promoted Prometheus and LD paths in vLLM Dockerfile; added cuda-python pin note and uv usage; support matrix footnote.
vLLM deployment resource placement
components/backends/vllm/deploy/agg_router.yaml
Moved GPU limit from container-level to pod-level resources.limits.gpu: "1".
Tests adjustments
tests/serve/test_sglang.py, tests/serve/test_vllm.py, tests/kvbm/test_determinism.py
Updated models in SGLang tests; relaxed one content-length assertion; increased vLLM readiness timeout 300→500; limited kvbm test markers (keep only kvbm).

Sequence Diagram(s)

sequenceDiagram
  autonumber
  participant Client
  participant Frontend (Dynamo)
  participant Worker (SGLang)

  rect rgb(245,248,255)
  note over Frontend (Dynamo): Tokenization enforced (skip_tokenizer_init=True)
  Client->>Frontend (Dynamo): /v1/chat/completions
  Frontend (Dynamo)->>Frontend (Dynamo): Tokenize input
  Frontend (Dynamo)->>Worker (SGLang): Prefill/Decode request (no tokenizer init)
  end

  alt Streaming tokens
    Worker (SGLang)-->>Frontend (Dynamo): res with output_ids
    Frontend (Dynamo)->>Client: stream tokens
  else Missing output_ids
    Worker (SGLang)-->>Frontend (Dynamo): res without output_ids
    Frontend (Dynamo)->>Frontend (Dynamo): raise ValueError with keys hint
    Frontend (Dynamo)-->>Client: error response
  end

  Frontend (Dynamo)->>Frontend (Dynamo): Detokenize (if needed)
  Frontend (Dynamo)-->>Client: final response
Loading

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

Poem

In cables and crates I twitch my nose,
Pins hop forward as the UCX grows.
Qwen now purrs where DeepSeek sat,
Frontend nibbles tokens—imagine that!
Wheels roll rc6, the tests lie in wait—
A rabbit stamps OK: ship this update. 🐰✨

Tip

🔌 Remote MCP (Model Context Protocol) integration is now available!

Pro plan users can now connect to remote MCP servers from the Integrations page. Connect with popular remote MCPs such as Notion and Linear to add more context to your reviews and chats.


📜 Recent review details

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

💡 Knowledge Base configuration:

  • MCP integration is disabled by default for public repositories
  • Jira integration is disabled by default for public repositories
  • Linear integration is disabled by default for public repositories

You can enable these sources in your CodeRabbit configuration.

📥 Commits

Reviewing files that changed from the base of the PR and between e2e909f and 2d96f92.

⛔ Files ignored due to path filters (2)
  • Cargo.lock is excluded by !**/*.lock
  • lib/bindings/python/Cargo.lock is excluded by !**/*.lock
📒 Files selected for processing (45)
  • Cargo.toml (0 hunks)
  • README.md (2 hunks)
  • components/backends/sglang/README.md (1 hunks)
  • components/backends/sglang/deploy/agg.yaml (1 hunks)
  • components/backends/sglang/deploy/agg_router.yaml (1 hunks)
  • components/backends/sglang/deploy/disagg-multinode.yaml (1 hunks)
  • components/backends/sglang/deploy/disagg.yaml (2 hunks)
  • components/backends/sglang/deploy/disagg_planner.yaml (2 hunks)
  • components/backends/sglang/docs/multinode-examples.md (1 hunks)
  • components/backends/sglang/docs/sgl-hicache-example.md (3 hunks)
  • components/backends/sglang/launch/agg.sh (1 hunks)
  • components/backends/sglang/launch/agg_router.sh (2 hunks)
  • components/backends/sglang/launch/disagg.sh (2 hunks)
  • components/backends/sglang/slurm_jobs/scripts/h100.sh (1 hunks)
  • components/backends/sglang/slurm_jobs/scripts/worker_setup.py (1 hunks)
  • components/backends/sglang/src/dynamo/sglang/args.py (1 hunks)
  • components/backends/sglang/src/dynamo/sglang/request_handlers/decode_handler.py (1 hunks)
  • components/backends/trtllm/README.md (1 hunks)
  • components/backends/trtllm/deploy/README.md (0 hunks)
  • components/backends/trtllm/deploy/agg.yaml (1 hunks)
  • components/backends/trtllm/deploy/agg_router.yaml (1 hunks)
  • components/backends/trtllm/deploy/disagg.yaml (2 hunks)
  • components/backends/trtllm/deploy/disagg_router.yaml (2 hunks)
  • components/backends/trtllm/gemma3_sliding_window_attention.md (1 hunks)
  • components/backends/trtllm/gpt-oss.md (1 hunks)
  • components/backends/trtllm/launch/agg.sh (1 hunks)
  • components/backends/trtllm/launch/agg_router.sh (1 hunks)
  • components/backends/trtllm/launch/disagg.sh (1 hunks)
  • components/backends/trtllm/launch/disagg_router.sh (1 hunks)
  • components/backends/vllm/deploy/agg_router.yaml (1 hunks)
  • container/Dockerfile (2 hunks)
  • container/Dockerfile.kvbm (1 hunks)
  • container/Dockerfile.sglang (1 hunks)
  • container/Dockerfile.sglang-wideep (2 hunks)
  • container/Dockerfile.trtllm (5 hunks)
  • container/Dockerfile.vllm (2 hunks)
  • container/build.sh (4 hunks)
  • docs/support_matrix.md (1 hunks)
  • lib/async-openai-macros/Cargo.toml (0 hunks)
  • lib/async-openai-macros/src/lib.rs (0 hunks)
  • lib/async-openai/Cargo.toml (1 hunks)
  • pyproject.toml (1 hunks)
  • tests/kvbm/test_determinism.py (1 hunks)
  • tests/serve/test_sglang.py (2 hunks)
  • tests/serve/test_vllm.py (1 hunks)

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
🪧 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.
    • 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.
  • 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 the files in the src/scheduler package and generate a class diagram using mermaid and a README in the markdown format.

Support

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

CodeRabbit Commands (Invoked using PR/Issue comments)

Type @coderabbitai help to get the list of available commands.

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.

Status, Documentation and Community

  • Visit our Status Page to check the current availability of CodeRabbit.
  • 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.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants