[TRTLLM-9603][feat] Enable ConfigurableMoE test in the CI

[xxi-nv]

Description

Motivation:
The MoE refactoring work is in progress and will last for several months. This means that the original MoE flow and the refactored flow will co - exist for several months.
To safeguard the newly refactored flow, we need to add several test cases.
In the future, however, there will be only the after - refactored flow.
To minimize the impact, I introduce a local conftest to ensure that the original test names remain unchanged.

Solution:

• Introduce a local conftest to customize the test case names, ensuring that the original test names remain the same.
• Add several necessary tests in the post - merge phase, which may take approximately 0.5 hours.
• In the future, once all MoE operations switch to the new ConfigurableMoE, the newly added cases will be deprecated.

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.

Details

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.

Summary by CodeRabbit

Release Notes

• New Features

  • Added runtime validation for configurable MOE backend to prevent execution issues during compilation.

• Tests

  • Expanded test coverage for MOE configurations across multiple quantization and backend variants.
  • Improved test ID formatting for better clarity and consistency in test result matching and reporting.

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

[coderabbitai]

📝 Walkthrough

Walkthrough

Changes introduce ENABLE_CONFIGURABLE_MOE environment variable propagation into tests via new parameters and patching mechanisms, add pytest conftest modules for MoE test ID formatting during collection, add guard logic in configurable_moe.py for backend attribute synchronization, and expand test coverage in l0_dgx_b200.yml with various MoE backend configurations.

Changes

Cohort / File(s)	Summary
Module Initialization Guard tensorrt_llm/_torch/modules/fused_moe/configurable_moe.py	Adds conditional guard in __init__ to only sync backend attributes if self.backend is not None; includes assertion warning when torch.compile is active to ensure backend presence during compilation.
Test Infrastructure — Pytest Conftest for Test ID Handling tests/integration/defs/accuracy/conftest.py, tests/unittest/_torch/modules/conftest.py	Introduces pytest hooks pytest_sessionstart and pytest_collection_modifyitems with helper _convert_clean_to_original_moe_test_id() to rewrite and normalize MoE-related test IDs during collection and display, removing extraneous leading/trailing dashes from parameterized test IDs.
Integration Test Parameterization & Environment Propagation tests/integration/defs/accuracy/test_llm_api_pytorch.py	Adds enable_configurable_moe parameter (values 0 or 1) and mocker to multiple test methods; introduces patch_mpi_pool_session_for_env() function to propagate ENABLE_CONFIGURABLE_MOE environment variable to MPI child processes; includes conditional skip logic for unsupported moe_backend values when ENABLE_CONFIGURABLE_MOE=1; wires moe_config parameter with MoeConfig(backend=moe_backend) into PyTorch test configurations.
Unit Test Parameterization & Environment Patching tests/unittest/_torch/modules/test_fused_moe.py	Adds enable_configurable_moe and mocker parameters to multiple test functions (nvfp4, w4a8_nvfp4_fp8, mxfp4_mxfp8, wfp4a16, w4afp8, fp8_blockwise_cutlass variants); patches ENABLE_CONFIGURABLE_MOE environment variable to "1" only when enable_configurable_moe==1 and backend is TRTLLM; includes skip checks when ENABLE_CONFIGURABLE_MOE=1 with unsupported backends.
Test List Configuration tests/integration/test_lists/test-db/l0_dgx_b200.yml	Adds multiple new MoE test cases across pre_merge conditions, including nvfp4 variants, w4a8_nvfp4_fp8, mxfp4/mxfp8, wfp4a16 configurations, and online_eplb test variants with configurable MoE backends (nvfp4_4gpus, DeepSeekV3Lite scenarios, w4_4gpus online_eplb).

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Areas requiring extra attention:

• Pytest hook logic correctness in conftest modules: verify _convert_clean_to_original_moe_test_id() string conversion correctly transforms test IDs and that hook ordering/timing is correct
• Environment variable propagation: ensure patch_mpi_pool_session_for_env() correctly injects env vars into MPI child processes and that patching logic is applied consistently across test variants
• Parameter consistency: validate that enable_configurable_moe and mocker parameters are added uniformly across all modified test functions and skip conditions are correctly gated
• Backend condition logic: verify skip logic correctly filters on TRTLLM backend support and environment variable patching respects backend types
• Test ID normalization: confirm dash-stripping logic in pytest hooks handles edge cases (empty parameters, already-wrapped IDs, non-target tests)

Pre-merge checks and finishing touches

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 47.06% which is insufficient. The required threshold is 80.00%.	You can run @coderabbitai generate docstrings to improve docstring coverage.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Title check	✅ Passed	Title clearly describes the main change: enabling ConfigurableMoE tests in CI. It's specific, concise, and directly reflects the primary objective of the changeset.
Description check	✅ Passed	Description provides clear motivation (MoE refactoring co-existence), solution (conftest customization and new tests), and future deprecation plan, with test coverage section present but unpopulated.

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• 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

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

[coderabbitai]

Actionable comments posted: 0

♻️ Duplicate comments (3)

tests/unittest/_torch/modules/test_fused_moe.py (3)

1534-1546: Consistent application of enable_configurable_moe pattern in w4a8_nvfp4_fp8

This block mirrors the earlier gating and env-patching logic for the NVFP4 test and looks consistent and correct for the W4A8 NVFP4 FP8 case as well. No additional issues beyond the (optional) helper extraction already mentioned.

---

1962-1976: Configurable MoE toggle for mxfp4/mxfp8 test is aligned with previous pattern

The added parametrization, skip guard, and ENABLE_CONFIGURABLE_MOE patching are consistent with test_fused_moe_nvfp4 and respect the TRTLLM-only constraint when the flag is enabled. Nothing further to fix here.

---

2237-2251: Correct use of enable_configurable_moe for WFP4A16 test

The new parameter, skip condition, and env patch are implemented in the same way as the other tests and correctly restrict ENABLE_CONFIGURABLE_MOE=1 to the TRTLLM backend. Implementation looks good.

🧹 Nitpick comments (4)

tensorrt_llm/_torch/modules/fused_moe/configurable_moe.py (1)

171-187: Clarify backend guard and torch.compile assertion semantics

The new block works but is a bit inconsistent with surrounding contracts:

• backend’s setter already rejects None, so the if self.backend is not None: guard is effectively redundant here unless something bypasses the setter.
• The comment mentions “backend is none during torch.compile”, but the assertion is only executed when backend is not None, and it asserts not torch.compiler.is_compiling(). The message text (“Backend should not be none if not in torch.compile”) also doesn’t match the condition being asserted.

To avoid confusion and make failures more actionable, consider either:

• Dropping the if self.backend is not None: guard and the assertion entirely if you don’t actually expect backend to be None here, or
• Restructuring to explicitly check the intended invariant, e.g.:
  • If self.backend is None, assert that you are in a compile context, or
  • If torch.compiler.is_compiling() is True, skip this sync block entirely.

This would better document the expected states and keep the assertion aligned with the comment and error message.

tests/unittest/_torch/modules/conftest.py (1)

34-110: Fused-MoE unittest ID cleanup helper is consistent and well-scoped

The per-module conftest for test_fused_moe.py mirrors the integration helper:

• _convert_clean_to_original_moe_test_id only adjusts IDs for that file and only when needed.
• pytest_sessionstart rewrites CLI args back to the original (dashed) form so collection still matches existing IDs.
• pytest_collection_modifyitems cleans up the “ugly” [-...] / -- patterns for display without affecting other tests.

This is a reasonable, localized workaround while enable_configurable_moe exists.

tests/integration/defs/accuracy/test_llm_api_pytorch.py (1)

16-21: MPIPoolSession env patch helper is a good, targeted way to propagate ENABLE_CONFIGURABLE_MOE

patch_mpi_pool_session_for_env cleanly overrides MpiPoolSession._start_mpi_pool to:

• Preserve the original behavior (MPIPoolExecutor(max_workers=self.n_workers, path=sys.path)).
• Add env=env_vars so worker processes see the desired environment values without relying on fragile initializer hacks.

Using the mocker fixture keeps the patch scoped to individual tests and avoids leaking into unrelated suites. Just ensure the CI environment is on a mpi4py version where MPIPoolExecutor(..., env=...) is supported (3.x+), otherwise this will raise a TypeError.

Also applies to: 23-45

tests/unittest/_torch/modules/test_fused_moe.py (1)

1363-1376: Configurable MoE gating + env patching logic is correct; consider centralizing the pattern

The skip guard (enable_configurable_moe == 1 and moe_backend != "TRTLLM") correctly prevents invalid backend/flag combinations, and mocker.patch.dict cleanly confines ENABLE_CONFIGURABLE_MOE changes to the test scope with a sane default of "0" otherwise. This matches the intended semantics.

Since the same guard + patch block is repeated in other tests, you might optionally factor it into a small helper (e.g., _patch_enable_configurable_moe(mocker, enable_configurable_moe, moe_backend)) to keep behavior centralized if the flag semantics evolve.

📜 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 21f2ba7 and 6a81949.

📒 Files selected for processing (6)

• tensorrt_llm/_torch/modules/fused_moe/configurable_moe.py (1 hunks)
• tests/integration/defs/accuracy/conftest.py (1 hunks)
• tests/integration/defs/accuracy/test_llm_api_pytorch.py (7 hunks)
• tests/integration/test_lists/test-db/l0_dgx_b200.yml (3 hunks)
• tests/unittest/_torch/modules/conftest.py (1 hunks)
• tests/unittest/_torch/modules/test_fused_moe.py (4 hunks)

🧰 Additional context used

📓 Path-based instructions (2)

**/*.py

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

**/*.py: The code developed for TensorRT-LLM should conform to Python 3.8+
Indent Python code with 4 spaces; do not use tabs
Always maintain the namespace when importing in Python, even if only one class or function from a module is used (e.g., use from package.subpackage import foo and then foo.SomeClass() instead of from package.subpackage.foo import SomeClass)
Python filenames should use snake_case (e.g., some_file.py)
Python class names should use PascalCase (e.g., class SomeClass)
Python function and method names should use snake_case (e.g., def my_awesome_function():)
Python local variable names should use snake_case, with prefix k for variable names that start with a number (e.g., k_99th_percentile = ...)
Python global variables should use upper snake_case with prefix G (e.g., G_MY_GLOBAL = ...)
Python constants should use upper snake_case (e.g., MY_CONSTANT = ...)
Avoid shadowing variables declared in an outer scope in Python
Initialize all externally visible members of a Python class in the constructor
For Python interfaces that may be used outside a file, prefer docstrings over comments
Python comments should be reserved for code within a function, or interfaces that are local to a file
Use Google style docstrings for Python classes and functions, which can be parsed by Sphinx
Python attributes and variables can be documented inline with type and description (e.g., self.x = 5 followed by """<type>: Description of 'x'""" )
Avoid using reflection in Python when functionality can be easily achieved without reflection
When using try-except blocks in Python, limit the except clause to the smallest set of specific errors possible instead of catching all exceptions
When using try-except blocks in Python to handle multiple possible variable types (duck-typing), keep the body of the try as small as possible and use the else block to implement the logic

Files:

• tensorrt_llm/_torch/modules/fused_moe/configurable_moe.py
• tests/unittest/_torch/modules/conftest.py
• tests/integration/defs/accuracy/conftest.py
• tests/unittest/_torch/modules/test_fused_moe.py
• tests/integration/defs/accuracy/test_llm_api_pytorch.py

**/*.{cpp,h,cu,py}

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

All TensorRT-LLM Open Source Software code files should contain an NVIDIA copyright header that includes the current year at the top

Files:

• tensorrt_llm/_torch/modules/fused_moe/configurable_moe.py
• tests/unittest/_torch/modules/conftest.py
• tests/integration/defs/accuracy/conftest.py
• tests/unittest/_torch/modules/test_fused_moe.py
• tests/integration/defs/accuracy/test_llm_api_pytorch.py

🧠 Learnings (9)

📓 Common learnings

Learnt from: djns99
Repo: NVIDIA/TensorRT-LLM PR: 6915
File: cpp/tensorrt_llm/kernels/cutlass_kernels/moe_gemm/moe_kernels.cu:4010-4012
Timestamp: 2025-08-14T23:23:27.449Z
Learning: For MOE (Mixture of Experts) code reviews in TensorRT-LLM, avoid repeatedly suggesting finalize fusion validation checks and safety assertions. The user djns99 has indicated these suggestions are repetitive and unwanted across multiple MOE-related changes.

📚 Learning: 2025-09-17T02:48:52.732Z

Learnt from: tongyuantongyu
Repo: NVIDIA/TensorRT-LLM PR: 7781
File: tests/integration/test_lists/waives.txt:313-313
Timestamp: 2025-09-17T02:48:52.732Z
Learning: In TensorRT-LLM, `tests/integration/test_lists/waives.txt` is specifically for waiving/skipping tests, while other test list files like those in `test-db/` and `qa/` directories are for different test execution contexts (pre-merge, post-merge, QA tests). The same test appearing in both waives.txt and execution list files is intentional - the test is part of test suites but will be skipped due to the waiver.

Applied to files:

• tests/integration/test_lists/test-db/l0_dgx_b200.yml

📚 Learning: 2025-09-09T09:40:45.658Z

Learnt from: fredricz-20070104
Repo: NVIDIA/TensorRT-LLM PR: 7645
File: tests/integration/test_lists/qa/llm_function_core.txt:648-648
Timestamp: 2025-09-09T09:40:45.658Z
Learning: In TensorRT-LLM test lists, it's common and intentional for the same test to appear in multiple test list files when they serve different purposes (e.g., llm_function_core.txt for comprehensive core functionality testing and llm_function_core_sanity.txt for quick sanity checks). This duplication allows tests to be run in different testing contexts.

Applied to files:

• tests/integration/test_lists/test-db/l0_dgx_b200.yml

📚 Learning: 2025-08-26T09:49:04.956Z

Learnt from: pengbowang-nv
Repo: NVIDIA/TensorRT-LLM PR: 7192
File: tests/integration/test_lists/test-db/l0_dgx_b200.yml:56-72
Timestamp: 2025-08-26T09:49:04.956Z
Learning: In TensorRT-LLM test configuration files, the test scheduling system handles wildcard matching with special rules that prevent duplicate test execution even when the same tests appear in multiple yaml files with overlapping GPU wildcards (e.g., "*b200*" and "*gb200*").

Applied to files:

• tests/integration/test_lists/test-db/l0_dgx_b200.yml

📚 Learning: 2025-07-28T17:06:08.621Z

Learnt from: moraxu
Repo: NVIDIA/TensorRT-LLM PR: 6303
File: tests/integration/test_lists/qa/examples_test_list.txt:494-494
Timestamp: 2025-07-28T17:06:08.621Z
Learning: In TensorRT-LLM testing, it's common to have both CLI flow tests (test_cli_flow.py) and PyTorch API tests (test_llm_api_pytorch.py) for the same model. These serve different purposes: CLI flow tests validate the traditional command-line workflow, while PyTorch API tests validate the newer LLM API backend. Both are legitimate and should coexist.

Applied to files:

• tests/integration/test_lists/test-db/l0_dgx_b200.yml

📚 Learning: 2025-08-14T23:23:27.449Z

Learnt from: djns99
Repo: NVIDIA/TensorRT-LLM PR: 6915
File: cpp/tensorrt_llm/kernels/cutlass_kernels/moe_gemm/moe_kernels.cu:4010-4012
Timestamp: 2025-08-14T23:23:27.449Z
Learning: For MOE (Mixture of Experts) code reviews in TensorRT-LLM, avoid repeatedly suggesting finalize fusion validation checks and safety assertions. The user djns99 has indicated these suggestions are repetitive and unwanted across multiple MOE-related changes.

Applied to files:

• tests/integration/test_lists/test-db/l0_dgx_b200.yml
• tests/unittest/_torch/modules/test_fused_moe.py

📚 Learning: 2025-08-06T13:58:07.506Z

Learnt from: galagam
Repo: NVIDIA/TensorRT-LLM PR: 6487
File: tests/unittest/_torch/auto_deploy/unit/singlegpu/test_ad_trtllm_bench.py:1-12
Timestamp: 2025-08-06T13:58:07.506Z
Learning: In TensorRT-LLM, test files (files under tests/ directories) do not require NVIDIA copyright headers, unlike production source code files. Test files typically start directly with imports, docstrings, or code.

Applied to files:

• tests/integration/test_lists/test-db/l0_dgx_b200.yml

📚 Learning: 2025-08-26T09:37:10.463Z

Learnt from: jiaganc
Repo: NVIDIA/TensorRT-LLM PR: 7031
File: tensorrt_llm/bench/dataclasses/configuration.py:90-104
Timestamp: 2025-08-26T09:37:10.463Z
Learning: In TensorRT-LLM's bench configuration, the `get_pytorch_perf_config()` method returns `self.pytorch_config` which is a Dict[str, Any] that can contain default values including `cuda_graph_config`, making the fallback `llm_args["cuda_graph_config"]` safe to use.

Applied to files:

• tests/integration/defs/accuracy/test_llm_api_pytorch.py

📚 Learning: 2025-08-26T09:37:10.463Z

Learnt from: jiaganc
Repo: NVIDIA/TensorRT-LLM PR: 7031
File: tensorrt_llm/bench/dataclasses/configuration.py:90-104
Timestamp: 2025-08-26T09:37:10.463Z
Learning: In TensorRT-LLM, the `get_pytorch_perf_config()` method returns `self.pytorch_config` which can contain default `cuda_graph_config` values, so `llm_args` may already have this config before the extra options processing.

Applied to files:

• tests/integration/defs/accuracy/test_llm_api_pytorch.py

🧬 Code graph analysis (4)

tensorrt_llm/_torch/modules/fused_moe/configurable_moe.py (1)

tensorrt_llm/llmapi/llm_args.py (2)

• slot_start (339-344)
• slot_end (347-349)

tests/unittest/_torch/modules/conftest.py (1)

tests/integration/defs/accuracy/conftest.py (3)

• _convert_clean_to_original_moe_test_id (33-58)
• pytest_sessionstart (61-70)
• pytest_collection_modifyitems (73-106)

tests/integration/defs/accuracy/conftest.py (1)

tests/unittest/_torch/modules/conftest.py (3)

• _convert_clean_to_original_moe_test_id (34-60)
• pytest_sessionstart (63-73)
• pytest_collection_modifyitems (76-110)

tests/integration/defs/accuracy/test_llm_api_pytorch.py (1)

tensorrt_llm/llmapi/mpi_session.py (1)

• MpiPoolSession (136-180)

⏰ 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 (3)

tests/integration/test_lists/test-db/l0_dgx_b200.yml (1)

20-23: Added ConfigurableMoE variants in B200 test matrix look consistent

The newly added entries for enable_configurable_moe variants (fused_moe unittests and DeepSeekV3Lite/GPTOSS accuracy tests) match the parametrizations introduced in the Python tests and stay within the existing l0_dgx_b200 conditions (4-GPU B200, pre/post-merge, PyTorch+MPI).

No structural or scheduling issues are apparent here; adding these IDs should simply expand the CI coverage for the ConfigurableMoE path.

Also applies to: 165-166, 200-202, 209-209

tests/integration/defs/accuracy/conftest.py (1)

34-106: Local MoE test ID normalization for integration tests looks sound

The _convert_clean_to_original_moe_test_id, pytest_sessionstart, and pytest_collection_modifyitems hooks correctly mirror the unittest behavior while scoping to test_llm_api_pytorch.py. They:

• Only rewrite IDs for the targeted file and bracketed params.
• Restore original IDs for collection while presenting cleaned IDs for reports/waives.
• Avoid touching non-MoE tests.

This should keep existing test list and waive references stable while allowing the new enable_configurable_moe parameterization to remain user-friendly.

tests/integration/defs/accuracy/test_llm_api_pytorch.py (1)

1874-1890: Configurable-MoE test parametrization and env wiring look correct and constrained

Across the DeepSeekV3Lite and GPTOSS tests you:

• Add an enable_configurable_moe ∈ {0,1} parameter with IDs "" for 0 and "enable_configurable_moe" for 1, keeping legacy test IDs stable while still exposing the new dimension.
• Guard enable_configurable_moe == 1 to only run when moe_backend == "TRTLLM" (where applicable), skipping unsupported backend combinations early with clear messages.
• Use patch_mpi_pool_session_for_env to set ENABLE_CONFIGURABLE_MOE to "1" only for the TRTLLM backend / enabled cases and to "0" otherwise, ensuring MPI workers don’t inherit stale values from the environment.

This gives you explicit coverage for the ConfigurableMoE path without disturbing existing test list naming or non-TRTLLM backends. The pattern is consistent across all modified tests.

Also applies to: 3515-3531, 3983-4031, 4343-4353

[xxi-nv]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #26705 [ run ] triggered by Bot. Commit: 563e595

[xxi-nv]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #26709 [ run ] triggered by Bot. Commit: dbdc279

[tensorrt-cicd]

PR_Github #26705 [ run ] completed with state ABORTED. Commit: 563e595
LLM/main/L0_MergeRequest_PR #20321 (Blue Ocean) completed with status: ABORTED

[syuoni]

@EmmaQiaoCh Could you please review this PR? It introduces some pytest utils, I'm wondering if we have some practice guidance on this.

[tensorrt-cicd]

PR_Github #26709 [ run ] completed with state SUCCESS. Commit: dbdc279
/LLM/main/L0_MergeRequest_PR pipeline #20326 completed with status: 'FAILURE'

[xxi-nv]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #26778 [ run ] triggered by Bot. Commit: f1179e6

[tensorrt-cicd]

PR_Github #26778 [ run ] completed with state SUCCESS. Commit: f1179e6
/LLM/main/L0_MergeRequest_PR pipeline #20389 completed with status: 'FAILURE'

[xxi-nv]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #26852 [ run ] triggered by Bot. Commit: f1179e6

[tensorrt-cicd]

PR_Github #26852 [ run ] completed with state SUCCESS. Commit: f1179e6
/LLM/main/L0_MergeRequest_PR pipeline #20453 completed with status: 'FAILURE'

[xxi-nv]

/bot run --disable-fail-fast

[xxi-nv]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #26909 [ run ] triggered by Bot. Commit: 7cf8d53

[tensorrt-cicd]

PR_Github #26909 [ run ] completed with state SUCCESS. Commit: 7cf8d53
/LLM/main/L0_MergeRequest_PR pipeline #20502 completed with status: 'FAILURE'

[xxi-nv]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #26963 [ run ] triggered by Bot. Commit: bae0866

[tensorrt-cicd]

PR_Github #26963 [ run ] completed with state SUCCESS. Commit: bae0866
/LLM/main/L0_MergeRequest_PR pipeline #20550 completed with status: 'FAILURE'

[xxi-nv]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #27028 [ run ] triggered by Bot. Commit: bae0866

[tensorrt-cicd]

PR_Github #27028 [ run ] completed with state SUCCESS. Commit: bae0866
/LLM/main/L0_MergeRequest_PR pipeline #20610 completed with status: 'FAILURE'

[xxi-nv]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #27166 [ run ] triggered by Bot. Commit: bae0866

[tensorrt-cicd]

PR_Github #27166 [ run ] completed with state FAILURE. Commit: bae0866
/LLM/main/L0_MergeRequest_PR pipeline #20729 completed with status: 'FAILURE'

[xxi-nv]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #27184 [ run ] triggered by Bot. Commit: bae0866

[tensorrt-cicd]

PR_Github #27184 [ run ] completed with state SUCCESS. Commit: bae0866
/LLM/main/L0_MergeRequest_PR pipeline #20746 completed with status: 'SUCCESS'

[HuiGao-NV]

LGTM