Skip to content

Conversation

@atchernych
Copy link
Contributor

@atchernych atchernych commented Aug 5, 2025

Overview:

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

    • Added detailed Kubernetes deployment documentation and templates for vLLM, SGLang, and TRTLLM backends, including multiple deployment patterns and configuration options.
    • Introduced advanced port allocation and management utilities for distributed deployments, improving reliability and flexibility.
  • Bug Fixes

    • Corrected and clarified various documentation links and command-line usage instructions across multiple READMEs and deployment guides.
    • Fixed package version constraints and dependencies to ensure compatibility, including explicit version pinning for Triton and NIXL.
  • Documentation

    • Significantly expanded and reorganized deployment and backend documentation, consolidating redundant sections and improving clarity for end-users.
    • Updated and streamlined Kubernetes deployment instructions, troubleshooting guides, and backend-specific configuration references.
  • Chores

    • Updated Dockerfiles and build scripts to use newer versions of dependencies and system utilities for improved compatibility and health monitoring.

alec-flowers and others added 24 commits July 30, 2025 22:07
Co-authored-by: Anant Sharma <anants@nvidia.com>
Co-authored-by: Ishan Dhanani <idhanani@nvidia.com>
Co-authored-by: Tanmay Verma <tanmayv@nvidia.com>
Co-authored-by: Dmitry Tokarev <dtokarev@nvidia.com>
Co-authored-by: Dmitry Tokarev <dtokarev@nvidia.com>
Co-authored-by: Dmitry Tokarev <dtokarev@nvidia.com>
…link (#2239)

Co-authored-by: Keiven Chang <keivenchang@users.noreply.github.com>
Signed-off-by: Anish <80174047+athreesh@users.noreply.github.com>
Co-authored-by: Anish <80174047+athreesh@users.noreply.github.com>
…2260)

Signed-off-by: Biswa Panda <biswa.panda@gmail.com>
@coderabbitai
Copy link
Contributor

coderabbitai bot commented Aug 5, 2025

Warning

Rate limit exceeded

@github-actions[bot] has exceeded the limit for the number of commits or files that can be reviewed per hour. Please wait 1 minutes and 17 seconds before requesting another review.

⌛ How to resolve this issue?

After the wait time has elapsed, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout.

Please see our FAQ for further information.

📥 Commits

Reviewing files that changed from the base of the PR and between 347620a and a871d03.

⛔ 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 (81)
  • README.md (3 hunks)
  • benchmarks/llm/README.md (0 hunks)
  • components/README.md (1 hunks)
  • components/backends/llama_cpp/README.md (1 hunks)
  • components/backends/sglang/README.md (3 hunks)
  • components/backends/sglang/deploy/README.md (1 hunks)
  • components/backends/sglang/deploy/disagg.yaml (2 hunks)
  • components/backends/sglang/docs/dsr1-wideep-h100.md (8 hunks)
  • components/backends/sglang/docs/multinode-examples.md (2 hunks)
  • components/backends/sglang/docs/sgl-http-server.md (1 hunks)
  • components/backends/sglang/launch/agg_router.sh (1 hunks)
  • components/backends/sglang/slurm_jobs/README.md (2 hunks)
  • components/backends/sglang/slurm_jobs/job_script_template.j2 (1 hunks)
  • components/backends/sglang/slurm_jobs/scripts/worker_setup.py (2 hunks)
  • components/backends/trtllm/README.md (4 hunks)
  • components/backends/trtllm/deploy/README.md (1 hunks)
  • components/backends/trtllm/deploy/agg.yaml (1 hunks)
  • components/backends/trtllm/deploy/agg_router.yaml (1 hunks)
  • components/backends/trtllm/deploy/disagg.yaml (1 hunks)
  • components/backends/trtllm/deploy/disagg_router.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/agg.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/decode.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/deepseek_r1/mtp/mtp_agg.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/deepseek_r1/mtp/mtp_decode.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/deepseek_r1/mtp/mtp_prefill.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/deepseek_r1/simple/agg.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/deepseek_r1/simple/decode.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/deepseek_r1/simple/prefill.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/deepseek_r1/wide_ep/dep16_agg.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/deepseek_r1/wide_ep/wide_ep_agg.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/deepseek_r1/wide_ep/wide_ep_decode.yaml (2 hunks)
  • components/backends/trtllm/engine_configs/deepseek_r1/wide_ep/wide_ep_prefill.yaml (2 hunks)
  • components/backends/trtllm/engine_configs/llama4/eagle/eagle_agg.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/llama4/eagle/eagle_decode.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/llama4/eagle/eagle_prefill.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/llama4/eagle_one_model/eagle_agg.yml (1 hunks)
  • components/backends/trtllm/engine_configs/llama4/eagle_one_model/eagle_decode.yaml (2 hunks)
  • components/backends/trtllm/engine_configs/llama4/eagle_one_model/eagle_prefill.yaml (1 hunks)
  • components/backends/trtllm/engine_configs/prefill.yaml (1 hunks)
  • components/backends/trtllm/src/dynamo/trtllm/main.py (1 hunks)
  • components/backends/vllm/README.md (3 hunks)
  • components/backends/vllm/deploy/README.md (1 hunks)
  • components/backends/vllm/src/dynamo/vllm/args.py (5 hunks)
  • components/backends/vllm/src/dynamo/vllm/ports.py (1 hunks)
  • container/Dockerfile.sglang (3 hunks)
  • container/Dockerfile.sglang-wideep (3 hunks)
  • container/Dockerfile.tensorrt_llm (4 hunks)
  • container/Dockerfile.vllm (2 hunks)
  • container/build.sh (2 hunks)
  • deploy/cloud/README.md (1 hunks)
  • deploy/cloud/helm/deploy.sh (1 hunks)
  • deploy/cloud/helm/dynamo-platform-values.yaml (0 hunks)
  • deploy/cloud/helm/platform/components/operator/templates/deployment.yaml (1 hunks)
  • deploy/cloud/helm/platform/components/operator/templates/manager-rbac.yaml (0 hunks)
  • deploy/cloud/helm/platform/components/operator/values.yaml (1 hunks)
  • deploy/cloud/helm/platform/values.yaml (1 hunks)
  • deploy/cloud/operator/cmd/main.go (6 hunks)
  • deploy/cloud/operator/internal/consts/consts.go (2 hunks)
  • deploy/cloud/operator/internal/controller/dynamographdeployment_controller.go (2 hunks)
  • deploy/cloud/operator/internal/controller_common/predicate.go (2 hunks)
  • deploy/cloud/operator/internal/dynamo/graph.go (2 hunks)
  • deploy/cloud/operator/internal/dynamo/graph_test.go (3 hunks)
  • deploy/inference-gateway/README.md (2 hunks)
  • deploy/metrics/README.md (1 hunks)
  • docs/API/nixl_connect/connector.md (1 hunks)
  • docs/architecture/dynamo_flow.md (1 hunks)
  • docs/components/backends/llm/README.md (0 hunks)
  • docs/examples/README.md (1 hunks)
  • docs/guides/dynamo_deploy/README.md (1 hunks)
  • docs/guides/dynamo_deploy/operator_deployment.md (0 hunks)
  • docs/guides/dynamo_deploy/quickstart.md (1 hunks)
  • docs/guides/dynamo_run.md (2 hunks)
  • docs/runtime/README.md (1 hunks)
  • examples/README.md (2 hunks)
  • examples/basics/disaggregated_serving/README.md (1 hunks)
  • examples/basics/multimodal/README.md (0 hunks)
  • examples/basics/multinode/README.md (1 hunks)
  • examples/basics/quickstart/README.md (1 hunks)
  • examples/deployments/EKS/Deploy_VLLM_example.md (1 hunks)
  • lib/llm/Cargo.toml (1 hunks)
  • pyproject.toml (1 hunks)

Walkthrough

This update extensively revises documentation, deployment guides, and configuration files across the project, especially for backend deployment patterns (vLLM, SGLang, TRTLLM) and Kubernetes integration. It introduces new deployment README files, restructures configuration YAMLs, updates Dockerfiles for dependency management, and removes or consolidates redundant documentation. Helm charts and operator logic are updated to replace a boolean Grove flag with a termination delay parameter and runtime Grove detection. Port allocation for vLLM is modularized into a new utility module.

Changes

Cohort / File(s) Change Summary
Documentation Restructuring and Cleanup
README.md, examples/README.md, docs/examples/README.md, docs/guides/dynamo_deploy/README.md, docs/guides/dynamo_deploy/operator_deployment.md, docs/components/backends/llm/README.md, examples/basics/multimodal/README.md
Reorganized and expanded framework support and deployment instructions; removed redundant or outdated documentation including operator deployment and multimodal example READMEs.
Backend Deployment Guides (Kubernetes)
components/backends/vllm/deploy/README.md, components/backends/sglang/deploy/README.md, components/backends/trtllm/deploy/README.md
Added comprehensive deployment guides for vLLM, SGLang, and TRTLLM backends detailing CRD structure, deployment patterns, configuration, and troubleshooting.
Kubernetes Deployment YAMLs
components/backends/trtllm/deploy/agg.yaml, components/backends/trtllm/deploy/agg_router.yaml, components/backends/trtllm/deploy/disagg.yaml, components/backends/trtllm/deploy/disagg_router.yaml
Introduced new DynamoGraphDeployment CR YAMLs for TRTLLM backend, defining various deployment patterns with health checks, resource configs, and container specs.
SGLang, vLLM, TRTLLM Backend Docs and Examples
components/backends/sglang/README.md, components/backends/trtllm/README.md, components/backends/vllm/README.md, components/backends/sglang/docs/dsr1-wideep-h100.md, components/backends/sglang/docs/multinode-examples.md, components/backends/sglang/docs/sgl-http-server.md, components/backends/sglang/launch/agg_router.sh, components/backends/sglang/slurm_jobs/README.md, components/backends/sglang/slurm_jobs/job_script_template.j2, components/backends/sglang/slurm_jobs/scripts/worker_setup.py
Updated documentation links, deployment instructions, and command-line examples; clarified multi-node and slurm job setup; restructured example references and removed broken/duplicate links.
Configuration File Restructuring (TRTLLM)
components/backends/trtllm/engine_configs/*
Refactored YAML config files: consolidated CUDA graph and cache config into nested structures, added dtype and cache transceiver config, and restructured MoE backend settings.
Port Allocation Refactor (vLLM)
components/backends/vllm/src/dynamo/vllm/args.py, components/backends/vllm/src/dynamo/vllm/ports.py
Modularized and centralized port allocation logic into a new ports.py utility; updated CLI args and config handling for port ranges and ETCD-based port reservation.
Dockerfile and Build Script Updates
container/Dockerfile.sglang, container/Dockerfile.sglang-wideep, container/Dockerfile.tensorrt_llm, container/Dockerfile.vllm, container/build.sh
Updated NIXL and TensorRT-LLM versions, added system dependencies (e.g., jq), locked triton version, clarified build steps, and improved comments for clarity and compatibility.
Helm Chart and Operator Grove Refactor
deploy/cloud/helm/deploy.sh, deploy/cloud/helm/dynamo-platform-values.yaml, deploy/cloud/helm/platform/components/operator/templates/deployment.yaml, deploy/cloud/helm/platform/components/operator/templates/manager-rbac.yaml, deploy/cloud/helm/platform/components/operator/values.yaml, deploy/cloud/helm/platform/values.yaml, deploy/cloud/operator/cmd/main.go, deploy/cloud/operator/internal/consts/consts.go, deploy/cloud/operator/internal/controller/dynamographdeployment_controller.go, deploy/cloud/operator/internal/controller_common/predicate.go, deploy/cloud/operator/internal/dynamo/graph.go, deploy/cloud/operator/internal/dynamo/graph_test.go
Replaced enableGrove boolean with groveTerminationDelay parameter; implemented runtime Grove detection; updated RBAC and deployment templates; added test and constant for termination delay.
Miscellaneous Documentation and Config Updates
benchmarks/llm/README.md, components/README.md, components/backends/llama_cpp/README.md, deploy/cloud/README.md, deploy/inference-gateway/README.md, deploy/metrics/README.md, docs/API/nixl_connect/connector.md, docs/architecture/dynamo_flow.md, docs/guides/dynamo_deploy/quickstart.md, docs/guides/dynamo_run.md, docs/runtime/README.md, examples/basics/disaggregated_serving/README.md, examples/basics/multinode/README.md, examples/basics/quickstart/README.md, examples/deployments/EKS/Deploy_VLLM_example.md, lib/llm/Cargo.toml, pyproject.toml, components/backends/trtllm/src/dynamo/trtllm/main.py
Corrected links, updated dependency constraints, fixed minor typos, improved documentation clarity, and made minor code/config fixes.

Sequence Diagram(s)

sequenceDiagram
    participant User
    participant Docs
    participant K8s
    participant Backend (vLLM/SGLang/TRTLLM)
    participant Operator
    participant ETCD

    User->>Docs: Refer to deployment guide (README)
    User->>K8s: Apply CR YAML (DynamoGraphDeployment)
    K8s->>Operator: Notify of new CR
    Operator->>K8s: Create Frontend/Worker Pods
    Backend->>ETCD: Allocate/reserve ports (if needed)
    User->>K8s: Port-forward to Frontend
    User->>Backend: Send inference request via Frontend
    Backend-->>User: Return inference result
Loading

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~45 minutes

Possibly related PRs

Poem

A bunny with docs in its paws,
Hopped through configs, YAMLs, and laws.
It tidied the charts,
Refactored the parts,
Now clusters deploy with applause!
🐇✨

(If you seek a backend, just follow the guide—
With Grove and ports, you’ll enjoy the ride!)


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.
    • 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.
  • 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.

Support

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

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.

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.