[https://nvbugs/5670793][fix] Solve trtllm-serve launch_disaggregated…

[JunyiXu-nv]

… port conflict

Summary by CodeRabbit

Release Notes

• Refactor
  • Enhanced inter-process communication infrastructure for model serving. The implementation now uses an optimized address allocation mechanism for internal process coordination, improving resource efficiency and performance during deployment and execution.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

Description

Test Coverage

PR Checklist

Please review the following before submitting your PR:

• PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.

• PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.

• Test cases are provided for new code paths (see test instructions)

• Any new dependencies have been scanned for license and vulnerabilities

• CODEOWNERS updated if ownership changes

• Documentation updated as needed

• Update tava architecture diagram if there is a significant design change in PR.

• The reviewers assigned automatically/manually are appropriate for the PR.

• Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...

Provide a user friendly way for developers to interact with a Jenkins server.

Run /bot [-h|--help] to print this help message.

See details below for each supported subcommand.

run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]

Launch build/test pipelines. All previously running jobs will be killed.

--reuse-test (optional)pipeline-id (OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.

--disable-reuse-test (OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.

--disable-fail-fast (OPTIONAL) : Disable fail fast on build/tests/infra failures.

--skip-test (OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.

--stage-list "A10-PyTorch-1, xxx" (OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.

--gpu-type "A30, H100_PCIe" (OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.

--test-backend "pytorch, cpp" (OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.

--only-multi-gpu-test (OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.

--disable-multi-gpu-test (OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.

--add-multi-gpu-test (OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.

--post-merge (OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.

--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" (OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".

--detailed-log (OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.

--debug (OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in the stage-list parameter to access the appropriate container environment. Note: Does NOT update GitHub check status.

For guidance on mapping tests to stage names, see docs/source/reference/ci-overview.md
and the scripts/test_to_stage_mapping.py helper.

kill

kill

Kill all running builds associated with pull request.

skip

skip --comment COMMENT

Skip testing for latest commit on pull request. --comment "Reason for skipping build/test" is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

reuse-pipeline

reuse-pipeline

Reuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

[JunyiXu-nv]

/bot run

[tensorrt-cicd]

PR_Github #25143 [ run ] triggered by Bot. Commit: 760dfc4

[coderabbitai]

📝 Walkthrough

Walkthrough

The pull request refactors inter-process communication in the TensorRT-LLM serving infrastructure by replacing TCP port allocation with IPC address generation. A new utility function generates IPC addresses using temporary file paths with UUIDs, and the disaggregated leader process spawning is updated to use this mechanism instead of TCP ports.

Changes

Cohort / File(s)	Summary
IPC Address Utility tensorrt_llm/llmapi/mpi_session.py	Added new public function find_free_ipc_addr() that generates IPC addresses in the format ipc://<temp_dir>/rpc_<uuid> using temporary file paths and UUIDs.
Serve Command Integration tensorrt_llm/commands/serve.py	Updated import to use find_free_ipc_addr instead of find_free_port. Modified _launch_disaggregated_leader to set the TLLM_SPAWN_PROXY_PROCESS_IPC_ADDR environment variable with an IPC address instead of a TCP port string.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~5 minutes

• Simple utility function addition with straightforward UUID-based IPC address generation
• Straightforward import and environment variable substitution in existing code
• No complex logic or control flow changes
• Consistent, localized modifications across two files

Pre-merge checks and finishing touches

❌ Failed checks (2 warnings)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	You can run @coderabbitai generate docstrings to improve docstring coverage.
Description check	⚠️ Warning	The PR description is largely incomplete. Only a fragment ('… port conflict') is provided with no explanation of the issue, solution, test coverage, or checklist items completed.	Add a clear description of the port conflict issue being fixed and the IPC-based solution. Document which tests validate these changes. Complete the PR checklist items.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Title check	✅ Passed	The PR title clearly references the NVBugs ticket and indicates a fix for a port conflict issue in trtllm-serve launch_disaggregated, matching the code changes made.

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

Tip

📝 Customizable high-level summaries are now available in beta!

You can now customize how CodeRabbit generates the high-level summary in your pull requests — including its content, structure, tone, and formatting.

• Provide your own instructions using the high_level_summary_instructions setting.
• Format the summary however you like (bullet lists, tables, multi-section layouts, contributor stats, etc.).
• Use high_level_summary_in_walkthrough to move the summary from the description to the walkthrough section.

Example instruction:

"Divide the high-level summary into five sections:

1. 📝 Description — Summarize the main change in 50–60 words, explaining what was done.
2. 📓 References — List relevant issues, discussions, documentation, or related PRs.
3. 📦 Dependencies & Requirements — Mention any new/updated dependencies, environment variable changes, or configuration updates.
4. 📊 Contributor Summary — Include a Markdown table showing contributions:
   | Contributor | Lines Added | Lines Removed | Files Changed |
5. ✔️ Additional Notes — Add any extra reviewer context.
   Keep each section concise (under 200 words) and use bullet or numbered lists for clarity."

Note: This feature is currently in beta for Pro-tier users, and pricing will be announced later.

---

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

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 0

🧹 Nitpick comments (2)

tensorrt_llm/llmapi/mpi_session.py (1)

547-551: Remove redundant os import inside function.

The os module is already imported at the module level (line 3), so the local import here is redundant.

Apply this diff to use the existing module-level import:

 def find_free_ipc_addr() -> str:
-    import os
     import tempfile
     import uuid
     return f'ipc://{os.path.join(tempfile.gettempdir(), "rpc_" + str(uuid.uuid4()))}'

tensorrt_llm/commands/serve.py (1)

643-643: Consider updating the TODO comment.

The TODO comment about making port allocation atomic is less relevant now that the implementation uses IPC addresses with UUIDs, which largely avoids the race condition concern that exists with TCP port allocation.

📜 Review details

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 2cde4e4 and 760dfc4.

📒 Files selected for processing (2)

• tensorrt_llm/commands/serve.py (2 hunks)
• tensorrt_llm/llmapi/mpi_session.py (1 hunks)

🧰 Additional context used

🧠 Learnings (1)

📓 Common learnings

Learnt from: pcastonguay
Repo: NVIDIA/TensorRT-LLM PR: 7455
File: tensorrt_llm/_torch/pyexecutor/py_executor.py:1852-1860
Timestamp: 2025-09-02T13:42:44.885Z
Learning: In MPI communication within TensorRT-LLM pipeline parallelism, different communication types (tokens, logits, termination sync) must use disjoint tag namespaces to avoid message routing collisions when using the same source/destination patterns.

🧬 Code graph analysis (1)

tensorrt_llm/commands/serve.py (2)

tensorrt_llm/llmapi/mpi_session.py (1)

• find_free_ipc_addr (547-551)

tensorrt_llm/executor/utils.py (1)

• LlmLauncherEnvs (22-29)

⏰ 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: Pre-commit Check

🔇 Additional comments (2)

tensorrt_llm/commands/serve.py (1)

644-647: IPC transport compatibility verification complete—no issues found.

The code correctly supports ZeroMQ IPC transport with PAIR sockets:

• Address format: find_free_ipc_addr() generates valid ipc:// addresses with proper tempdir paths (mpi_session.py:551)
• Socket operations: ZeroMqQueue.socket.bind() and socket.connect() are generic operations that work with any ZeroMQ transport, including ipc:// (ipc.py:78-79, 103)
• PAIR socket support: ZeroMQ PAIR sockets natively support IPC transport (confirmed: zmq.PAIR used at ipc.py:34, mpi_session.py:424)
• Address passing: The address flows correctly from find_free_ipc_addr() through environment variables to RemoteMpiCommSessionServer and RemoteMpiCommSessionClient, then to ZeroMqQueue for bind/connect operations
• Error handling: After binding, getsockopt(zmq.LAST_ENDPOINT) properly retrieves the actual bound endpoint (ipc.py:81-82)

The switch from TCP to IPC is fully supported by the existing implementation and requires no changes.

tensorrt_llm/llmapi/mpi_session.py (1)

547-551: Verify whether Windows support is required for this deployment path.

The socket cleanup concern is properly addressed—ZeroMQ's context lifecycle (via context.term() in executor/ipc.py) automatically manages socket creation and cleanup. However, a platform compatibility issue exists: the code unconditionally uses the ipc:// protocol, which is unsupported on Windows (only works with Unix domain sockets on Linux, macOS, BSD).

No platform-specific handling or fallback exists in find_free_ipc_addr() or the serve.py launch chain. If Windows support is required, this function should either detect the platform and use tcp:// as a fallback on Windows, or document that Unix-like systems are required.

[tensorrt-cicd]

PR_Github #25143 [ run ] completed with state SUCCESS. Commit: 760dfc4
/LLM/release-1.1/L0_MergeRequest_PR pipeline #506 completed with status: 'SUCCESS'