docs: organize sglang readme

[ishandhanani]

🧹🧹🧹

Summary by CodeRabbit

• Documentation
  • Major restructuring and expansion of the SGLang examples README, including a detailed feature matrix, quick start instructions, advanced examples, and deployment guidance.
  • Added new documentation for running DeepSeek-R1 with Wide Expert Parallelism (WideEP) on GB200 hardware.
  • Updated instructions for running DeepSeek-R1 with WideEP on H100s, clarifying image names and hardware requirements.
  • Introduced documentation for the SGLang HTTP server, outlining its REST API, discovery architecture, and usage.
  • Added an Apache 2.0 license header to the multinode examples documentation.
  • Minor formatting update in the main README.

[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

This update introduces significant improvements and expansions to the documentation in the examples/sglang directory. The main README is restructured with detailed instructions, feature matrices, and categorized examples. New documentation files are added for advanced deployment scenarios and features, while some existing docs are updated or licensed.

Changes

File(s)	Change Summary
README.md	Removed trailing newline after the last code block.
examples/sglang/README.md	Extensively restructured and expanded with feature matrix, quick start, categorized examples, and deployment details.
examples/sglang/docs/dsr1-wideep-gb200.md	Added new documentation for running DeepSeek-R1 with WideEP on GB200 hardware.
examples/sglang/docs/dsr1-wideep-h100.md	Updated title and Docker references to specify H100 hardware and new image/tag names.
examples/sglang/docs/multinode-examples.md	Added Apache 2.0 license header; no content changes.
examples/sglang/docs/sgl-http-server.md	Added new documentation describing the SGLang HTTP server, its architecture, endpoints, and usage.

Sequence Diagram(s)

sequenceDiagram
    participant User
    participant HTTP_Server
    participant Etcd
    participant SGLang_Worker

    User->>HTTP_Server: POST /flush_cache
    HTTP_Server->>Etcd: Query for registered SGLang components
    Etcd-->>HTTP_Server: List of component endpoints
    loop For each component
        HTTP_Server->>SGLang_Worker: Send flush_cache request
        SGLang_Worker-->>HTTP_Server: Acknowledge flush
    end
    HTTP_Server-->>User: Return success status

Loading

Possibly related PRs

• chore(sglang): readme and instruction fixes #1761: Updates and fixes to the examples/sglang/README.md for deployment instructions, closely related to this PR's major restructuring and expansion of the same file.

Poem

In docs and diagrams, we hop anew,
With features mapped and guidance true.
WideEP on GB200s, H100s too,
HTTP servers flush as caches do.
🐰 With every markdown, clarity grows—
The Dynamo rabbit surely knows!

---

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: 3

🧹 Nitpick comments (8)

README.md (1)

206-206: No-op change – consider squashing to reduce noise

Only a trailing-newline deletion was introduced. While harmless, commits that touch nothing but EOF whitespace bloat history and trigger unnecessary CI cycles. Unless there is a policy that enforces it, consider squashing this hunk away before merging.

examples/sglang/docs/dsr1-wideep-h100.md (2)

38-39: Minor: use explicit tag for reproducibility

Consider pinning the image tag instead of latest to guarantee users build against the exact revision that matches these docs, e.g.:

-docker build -f container/Dockerfile.sglang-wideep . -t dynamo-wideep --no-cache
+docker build -f container/Dockerfile.sglang-wideep . -t dynamo-wideep:0.4.8-post1 --no-cache

---

59-60: Nit: quote image in multi-line docker run for readability

Long command snippets become easier to copy-paste when the image reference is on a separate line with continuation char:

...
\  
  dynamo-wideep:latest

Totally optional.

examples/sglang/docs/sgl-http-server.md (1)

22-22: Tautology: “API interface”

API already stands for Application Programming Interface. Drop “interface” for conciseness:

-The SGLang HTTP server provides a REST API interface for managing ...
+The SGLang HTTP server provides a REST API for managing ...

examples/sglang/README.md (4)

67-68: Fix redundant wording in Quick-Start intro

“run all of our the common deployment patterns” has a duplicate article.
[diff]

-Below we provide a guide that lets you run all of our the common deployment patterns on a single node.
+Below we provide a guide that lets you run all of our common deployment patterns on a single node.

---

58-63: Disambiguate mixed status “✅/🚧” in WideEP row

Using two emojis in the same cell makes it hard to know the actual state. Split the hardware dimension into separate rows (e.g., “WideEP on H100” ✅, “WideEP on GB200” 🚧) or add a footnote to explain.

---

120-123: Consistent heading capitalisation

The previous sections use title-case (“Aggregated Serving”), but this heading is “Disaggregated serving”. Capitalise “Serving” for consistency.

-### Disaggregated serving
+### Disaggregated Serving

---

132-132: Correct typo “conjuction”

-... used in conjuction with NIXL ...
+... used in conjunction with NIXL ...

📜 Review details

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

📥 Commits

Reviewing files that changed from the base of the PR and between 6cdda03 and ca5e9bb.

📒 Files selected for processing (6)

• README.md (1 hunks)
• examples/sglang/README.md (4 hunks)
• examples/sglang/docs/dsr1-wideep-gb200.md (1 hunks)
• examples/sglang/docs/dsr1-wideep-h100.md (3 hunks)
• examples/sglang/docs/multinode-examples.md (1 hunks)
• examples/sglang/docs/sgl-http-server.md (1 hunks)

🧰 Additional context used

🧠 Learnings (3)

examples/sglang/docs/multinode-examples.md (1)

Learnt from: fsaady
PR: ai-dynamo/dynamo#1730
File: examples/sglang/slurm_jobs/job_script_template.j2:59-59
Timestamp: 2025-07-02T13:20:28.800Z
Learning: In the SLURM job script template at examples/sglang/slurm_jobs/job_script_template.j2, the `--total_nodes` parameter represents the total nodes per worker type (prefill or decode), not the total nodes in the entire cluster. Each worker type needs to know its own group size for distributed coordination.

examples/sglang/docs/dsr1-wideep-h100.md (3)

Learnt from: fsaady
PR: ai-dynamo/dynamo#1730
File: examples/sglang/slurm_jobs/job_script_template.j2:59-59
Timestamp: 2025-07-02T13:20:28.800Z
Learning: In the SLURM job script template at examples/sglang/slurm_jobs/job_script_template.j2, the `--total_nodes` parameter represents the total nodes per worker type (prefill or decode), not the total nodes in the entire cluster. Each worker type needs to know its own group size for distributed coordination.

Learnt from: nnshah1
PR: ai-dynamo/dynamo#1444
File: tests/fault_tolerance/configs/agg_tp_1_dp_8.yaml:31-38
Timestamp: 2025-07-01T15:33:53.262Z
Learning: In fault tolerance test configurations, the `resources` section under `ServiceArgs` specifies resources per individual worker, not total resources for all workers. So `workers: 8` with `gpu: '1'` means 8 workers × 1 GPU each = 8 GPUs total.

Learnt from: julienmancuso
PR: ai-dynamo/dynamo#1474
File: deploy/cloud/operator/internal/controller/dynamocomponent_controller.go:1302-1306
Timestamp: 2025-06-11T21:18:00.425Z
Learning: In the Dynamo operator, the project’s preferred security posture is to set a Pod-level `PodSecurityContext` with `runAsUser`, `runAsGroup`, and `fsGroup` all set to `1000`, and then selectively override the user at the individual container level (e.g., `RunAsUser: 0` for Kaniko) when root is required.

examples/sglang/README.md (1)

Learnt from: fsaady
PR: ai-dynamo/dynamo#1730
File: examples/sglang/slurm_jobs/scripts/worker_setup.py:230-244
Timestamp: 2025-07-03T10:14:30.570Z
Learning: In examples/sglang/slurm_jobs/scripts/worker_setup.py, background processes (like nats-server, etcd) are intentionally left running even if later processes fail. This design choice allows users to manually connect to nodes and debug issues without having to restart the entire SLURM job from scratch, providing operational flexibility for troubleshooting in cluster environments.

🪛 LanguageTool

examples/sglang/docs/dsr1-wideep-gb200.md

[grammar] ~20-~20: Ensure spelling is correct
Context: ... SGLang's implementation of wide expert parallelilsm and large scale P/D for DeepSeek-R1 on ...

(QB_NEW_EN_ORTHOGRAPHY_ERROR_IDS_1)

examples/sglang/docs/sgl-http-server.md

[style] ~22-~22: This phrase is redundant (‘I’ stands for ‘interface’). Use simply “API”.
Context: ... The SGLang HTTP server provides a REST API interface for managing and monitoring SGLang comp...

(ACRONYM_TAUTOLOGY)

🪛 GitHub Actions: Pre Merge Validation of (ai-dynamo/dynamo/refs/pull/1910/merge) by ishandhanani.

examples/sglang/docs/dsr1-wideep-gb200.md

[error] 21-21: pre-commit hook 'trailing-whitespace' failed. Trailing whitespace was found and automatically fixed. Please run 'pre-commit run --all-files' locally to apply fixes.

⏰ 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 (4)

examples/sglang/docs/multinode-examples.md (1)

1-16: 👍 License header addition looks good

The Apache-2.0 stub is correct and matches project convention.

examples/sglang/docs/dsr1-wideep-gb200.md (1)

24-26: Incomplete instruction list

Steps end at “2. Build the Dynamo container” but no commands follow. Add the missing build/run example or remove the enumeration to avoid confusing readers.

examples/sglang/docs/sgl-http-server.md (1)

83-87: Default values: highlight that port 9001 collides with kv-flush script

Earlier docs use curl -X POST http://...:9001/flush_cache. Explicitly stating the default here avoids confusion—good! Consider adding a note that changing the port requires adjusting the benchmarking scripts.

examples/sglang/README.md (1)

71-71: Broken markdown link to compose file?

The inline link [Docker Compose](../../deploy/metrics/docker-compose.yml) moves two levels up from examples/sglang. The actual path appears to be ../../deploy/metrics/... only one level up (examples → root). Verify the depth and fix if necessary to avoid 404s.

[biswapanda]

minor comment, LGTM