[https://nvbugs/5445466][fix] Eliminate race when loading HF dynamic modules

[chang-l]

Possibly related issue: huggingface/transformers#37492

Summary by CodeRabbit

• New Features

  • Adds a timeout-aware global lock around loading pretrained model configurations to serialize concurrent access (falls back with a warning on timeout).

• Bug Fixes

  • Reduces intermittent race conditions and errors when multiple processes load model configs, improving reliability.

• Tests

  • Re-enables two previously skipped multi-GPU DeepSeekR1 tests, expanding test coverage.

Description

Test Coverage

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.

[coderabbitai]

📝 Walkthrough

Walkthrough

Adds a module-level context manager config_file_lock(timeout: int = 10) to serialize pretrained config loading in tensorrt_llm/_torch/model_config.py; uses it around ModelConfig.from_pretrained, moves model_dir resolution inside the lock, and removes two SKIP waivers so two DeepSeekR1 multi-GPU tests will run.

Changes

Cohort / File(s)	Summary
Torch model config locking tensorrt_llm/_torch/model_config.py	Adds config_file_lock(timeout: int = 10) context manager using a global lock at HF_MODULES_CACHE/_remote_code.lock (via filelock); wraps ModelConfig.from_pretrained (including AutoConfig.from_pretrained) with the lock; moves resolution of model_dir (transformers.utils.hub.cached_file(...).parent) inside the lock; logs and proceeds if lock acquisition times out; adds contextlib, filelock, and HF_MODULES_CACHE imports and docstring.
Test waivers update tests/integration/test_lists/waives.txt	Removes two SKIP entries that previously waived accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_nvfp4_multi_gpus variants ([latency_trtllmgen], [latency]), enabling those tests.

Sequence Diagram(s)

sequenceDiagram
  autonumber
  participant Caller as Caller
  participant MC as ModelConfig.from_pretrained
  participant Lock as config_file_lock
  participant Hub as transformers.utils.hub
  participant HF as AutoConfig.from_pretrained

  Caller->>MC: from_pretrained(checkpoint_dir)
  MC->>Lock: enter(timeout=10)
  alt lock acquired
    MC->>Hub: cached_file(checkpoint_dir, "config.json")
    Hub-->>MC: config.json path
    MC->>MC: model_dir = parent(path)
    MC->>HF: AutoConfig.from_pretrained(checkpoint_dir)
    HF-->>MC: config
    MC->>MC: load hf_quant_config.json, dtypes.json from model_dir
  else timeout (no lock)
    MC->>MC: log warning and proceed without lock
    MC->>Hub: cached_file(...)
    MC->>HF: AutoConfig.from_pretrained(...)
  end
  MC->>Lock: exit
  MC-->>Caller: ModelConfig instance

Loading

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

• [None][infra] Waive failed cases on main 08/14 #6902 — Modifies the same waives/test skip entries for DeepSeekR1 multi-GPU tests.

Suggested labels

weekly_release_blocker

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 f016d97 and 66f5943.

📒 Files selected for processing (1)

• tests/integration/test_lists/waives.txt (0 hunks)

💤 Files with no reviewable changes (1)

• tests/integration/test_lists/waives.txt

⏰ 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

✨ Finishing Touches

🧪 Generate unit tests

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

---

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.
  • Open a follow-up GitHub issue for this discussion.
• Files and specific lines of code (under the "Files changed" tab): Tag @coderabbit in a new review comment at the desired location with your query.
• PR comments: Tag @coderabbit 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:
  • @coderabbit gather interesting stats about this repository and render them as a table. Additionally, render a pie chart showing the language distribution in the codebase.
  • @coderabbit 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 @coderabbit help to get the list of available commands.

Other keywords and placeholders

• Add @coderabbit ignore or @coderabbitai ignore anywhere in the PR description to prevent this PR from being reviewed.
• Add @coderabbit summary or @coderabbitai summary to generate the high-level summary at a specific location in the PR description.
• Add @coderabbit or @coderabbitai title 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.

[chang-l]

/bot run --stage-list "GB200-8_GPUs-2_Nodes-PyTorch-Post-Merge-4"

[tensorrt-cicd]

PR_Github #16592 [ run ] triggered by Bot

[coderabbitai]

Actionable comments posted: 0

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

tensorrt_llm/_torch/model_config.py (1)

1-19: Missing filelock Dependency Declaration

The verification shows that filelock is not listed in any of the project’s dependency manifests (pyproject.toml, setup.py, requirements-dev.txt, requirements.txt). Without declaring it, installs will fail at runtime when importing the module.

• Add filelock (with an appropriate version specifier) to your project’s dependencies.
– If you’re using Poetry, include it under [tool.poetry.dependencies] in pyproject.toml.
– If you’re using setup.py, add it to the install_requires list.
– If you manage requirements via requirements.txt, append a line such as:
text filelock>=3.12.0

Once filelock is declared, CI installs and runtime imports will resolve correctly.

🧹 Nitpick comments (2)

tensorrt_llm/_torch/model_config.py (2)

64-91: Make the lock robust on first-run/NFS and reduce contention scope if desired

The locking approach is sound and addresses the HF dynamic modules race. A few practical robustness tweaks will make it safer in shared caches and CI:

• Ensure the lock directory exists before creating the lock file. On a fresh environment, Path(HF_MODULES_CACHE) may not exist yet.
• Allow opting into SoftFileLock for NFS/shared filesystems (common in multi-node jobs), gated by an env var to avoid behavior changes by default.
• Improve the warning to include the lock path and PID for diagnosability.

Apply this diff within the context manager:

 @contextlib.contextmanager
 def config_file_lock(timeout: int = 10):
     """
     Context manager for file locking when loading pretrained configs.

     This prevents race conditions when multiple processes try to download/load
     the same model configuration simultaneously.

     Args:
         timeout: Maximum time to wait for lock acquisition in seconds
     """
-    # Use a single global lock file in HF cache directory
-    # This serializes all model loading operations to prevent race conditions
-    lock_path = Path(HF_MODULES_CACHE) / "_remote_code.lock"
+    # Use a single global lock file in HF cache directory
+    # This serializes all model loading operations to prevent race conditions
+    lock_path = Path(transformers.utils.HF_MODULES_CACHE) / "_remote_code.lock"
+    # Ensure the directory exists (first-run safety)
+    lock_path.parent.mkdir(parents=True, exist_ok=True)

-    # Create and acquire the lock
-    lock = filelock.FileLock(str(lock_path), timeout=timeout)
+    # Create and acquire the lock. Allow opting into SoftFileLock for NFS via env var.
+    use_soft = os.environ.get("TRTLLM_CONFIG_LOCK_SOFT", "0") == "1"
+    lock_cls = filelock.SoftFileLock if use_soft else filelock.FileLock
+    lock = lock_cls(str(lock_path), timeout=timeout)

     try:
         with lock:
             yield
     except filelock.Timeout:
-        logger.warning(
-            f"Failed to acquire config lock within {timeout} seconds, proceeding without lock"
-        )
+        logger.warning(
+            f"Config lock timeout after {timeout}s at path={lock_path} pid={os.getpid()}, proceeding without lock"
+        )
         # Fallback: proceed without locking to avoid blocking indefinitely
         yield

---

393-406: Consider per-checkpoint locking and forwarding all HF kwargs

Both refinements remain optional but will make this loader more flexible under concurrency and with advanced Hugging Face options:

• Per‐checkpoint lock: you’re currently using a global lock around

  with config_file_lock():
      transformers.AutoConfig.from_pretrained(…)
      transformers.utils.hub.cached_file(…)

  For high‐throughput scenarios loading different models in parallel, you could key the lock on checkpoint_dir (or a hash thereof) so that unrelated model loads don’t block each other.
• Full HF kwargs support: in
  tensorrt_llm/_torch/models/checkpoints/hf/config_loader.py:11,
  you forward **kwargs into ModelConfig.from_pretrained(checkpoint_dir, **kwargs). However, the implementation in tensorrt_llm/_torch/model_config.py currently only accepts and forwards trust_remote_code to AutoConfig.from_pretrained (and uses only checkpoint_dir for cached_file). If you ever need to load specific revisions, use a custom cache directory, pass auth tokens, or operate offline, you’ll want to extend ModelConfig.from_pretrained to accept the usual HF parameters—revision, cache_dir, use_auth_token, local_files_only, etc.—and pass them through to both AutoConfig.from_pretrained and hub.cached_file.

No existing call sites pass extra HF kwargs today, but updating the signature now avoids future breakage as you onboard more advanced loading scenarios.

📜 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 87d1d3a and 17f884f.

📒 Files selected for processing (2)

• tensorrt_llm/_torch/model_config.py (3 hunks)
• tests/integration/test_lists/waives.txt (0 hunks)

💤 Files with no reviewable changes (1)

• tests/integration/test_lists/waives.txt

🧰 Additional context used

📓 Path-based instructions (2)

**/*.py

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

**/*.py: Code must target Python 3.8+
Indent Python code with 4 spaces; do not use tabs
Preserve module namespaces when importing; import modules/packages and access members via the module (e.g., from package.subpackage import foo; foo.SomeClass())
Python file names should be snake_case
Python class names should be PascalCase
Python functions/methods and local variables should be snake_case; variables beginning with a number should be prefixed with k_ (e.g., k_99th_percentile)
Global variables should be UPPER_SNAKE_CASE prefixed with G_ (e.g., G_MY_GLOBAL); constants should be UPPER_SNAKE_CASE
Avoid shadowing variables from outer scopes; initialize all externally visible members in init
Prefer docstrings for interfaces used outside a file; comments should be reserved for in-function or file-local interfaces
Use Google-style docstrings for classes and functions; attributes and variables may be documented inline with trailing string literals
Avoid reflection when simpler, explicit code suffices (e.g., avoid dict(**locals()) patterns)
In try/except, catch the narrowest exceptions possible
For duck-typing patterns, keep the try body minimal and move logic to else to avoid masking unrelated failures

Files:

• tensorrt_llm/_torch/model_config.py

**/*.{c,cc,cpp,cxx,h,hh,hpp,hxx,cu,cuh,py}

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

Prepend the NVIDIA copyright header (current year) to all source files (.cpp, .h, .cu, .py, etc.)

Files:

• tensorrt_llm/_torch/model_config.py

🧬 Code graph analysis (1)

tensorrt_llm/_torch/model_config.py (2)

tensorrt_llm/logger.py (1)

• warning (131-132)

tensorrt_llm/models/automodel.py (1)

• AutoConfig (10-49)

⏰ 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

[tensorrt-cicd]

PR_Github #16592 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #12457 (Partly Tested) completed with status: 'SUCCESS'

[chang-l]

/bot run --stage-list "GB200-8_GPUs-2_Nodes-PyTorch-Post-Merge-4"

[tensorrt-cicd]

PR_Github #16598 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #16598 [ run ] completed with state FAILURE
/LLM/main/L0_MergeRequest_PR pipeline #12461 (Partly Tested) completed with status: 'FAILURE'

[chang-l]

/bot run --stage-list "GB200-8_GPUs-2_Nodes-PyTorch-Post-Merge-3"

[tensorrt-cicd]

PR_Github #16632 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #16632 [ run ] completed with state ABORTED

[chang-l]

/bot run --stage-list "GB200-8_GPUs-2_Nodes-PyTorch-Post-Merge-3"

[chang-l]

/bot run

[tensorrt-cicd]

PR_Github #16789 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #16789 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #12605 completed with status: 'FAILURE'

[chang-l]

/bot run

[tensorrt-cicd]

PR_Github #16888 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #16888 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #12685 completed with status: 'SUCCESS'