[None][infra] Waive 12 failed cases for main in post-merge 2889

[trtllm-agent]

Auto-generated Waive PR

Created by: TensorRT LLM CI Report (requested by @mzweilz)
Target branch: main
Bug(s): 6572800, 6572835, 6572843

Waive entries added

examples/visual_gen/test_visual_gen_cosmos3.py::test_cosmos3_feature_accuracy_against_golden[nvfp4] SKIP (https://nvbugs/6572800)
examples/visual_gen/test_visual_gen_flux.py::test_flux_accuracy_against_golden[flux1-nvfp4] SKIP (https://nvbugs/6572800)
examples/visual_gen/test_visual_gen_flux.py::test_flux_accuracy_against_golden[flux2-nvfp4] SKIP (https://nvbugs/6572800)
examples/visual_gen/test_visual_gen_ltx2.py::test_ltx2_feature_accuracy_against_golden[nvfp4] SKIP (https://nvbugs/6572800)
examples/visual_gen/test_visual_gen_qwen_image.py::test_qwenimage_feature_accuracy_against_golden[cuda-graph] SKIP (https://nvbugs/6572800)
examples/visual_gen/test_visual_gen_qwen_image.py::test_qwenimage_feature_accuracy_against_golden[nvfp4] SKIP (https://nvbugs/6572800)
examples/visual_gen/test_visual_gen_wan.py::test_wan_feature_accuracy_against_golden[wan21-nvfp4] SKIP (https://nvbugs/6572800)
examples/visual_gen/test_visual_gen_wan.py::test_wan_feature_accuracy_against_golden[wan22-cuda-graph] SKIP (https://nvbugs/6572800)
examples/visual_gen/test_visual_gen_wan.py::test_wan_feature_accuracy_against_golden[wan22-fp8-blockwise] SKIP (https://nvbugs/6572800)
examples/visual_gen/test_visual_gen_wan.py::test_wan_feature_accuracy_against_golden[wan22-nvfp4] SKIP (https://nvbugs/6572800)
perf/test_perf_sanity.py::test_e2e[disagg_upload-e2e-gb300_deepseek-r1-fp4_128k8k_con256_ctx1_pp4_gen1_dep8_eplb0_mtp1_ccb-NIXL] SKIP (https://nvbugs/6572843)
unittest/_torch/modules/moe/test_moe_module.py::test_configurable_moe_single_gpu -k "MEGAMOE_CUTEDSL" SKIP (https://nvbugs/6572835)

---

This PR was auto-generated by TensorRT LLM CI Report. Please review the waive entries before merging.

Dev Engineer Review

• Added 12 waiver entries in tests/integration/test_lists/waives.txt.
• Entries cover visual-generation accuracy tests, one GB300 disaggregated performance test, and one single-GPU MegaMoE CUTEDSL test.
• Each entry references bug 6572800, 6572835, or 6572843.
• The change is additive and has no exported API changes.
• The waiver format and scope are consistent with the stated objectives.
• No duplicate or invalid entries were identified from the provided change summary.

QA Engineer Review

• Modified tests/integration/test_lists/waives.txt.
• Added 12 waiver entries. No entries were removed.
• The change affects test waivers only. It does not modify test implementations.
• CBTS coverage data is unavailable.
• Verdict: needs follow-up.

[mzweilz]

/bot skip --comment "waive only"

[coderabbitai]

Walkthrough

The integration waiver list now skips selected visual-generation accuracy tests, one GB300 DeepSeek-R1 FP4 disaggregated performance test, and one single-GPU MegaMoE CUTEDSL test. Each entry references an NVBug issue.

Changes

Integration test waivers

Layer / File(s)	Summary
Add integration test skip waivers tests/integration/test_lists/waives.txt	Adds skip entries for Cosmos3, Flux, LTX2, Qwen Image, Wan, GB300 DeepSeek-R1 FP4 disaggregated performance, and single-GPU MegaMoE CUTEDSL tests.

Estimated code review effort: 1 (Trivial) | ~3 minutes

Possibly related PRs

• NVIDIA/TensorRT-LLM#17402: Adds waivers for GB300 disaggregated performance tests.
• NVIDIA/TensorRT-LLM#17386: Adjusts GB300-related test waivers.
• NVIDIA/TensorRT-LLM#17384: Adds NVBug-linked skip entries for integration and performance tests.

Suggested reviewers: brnguyen2, qijune, bowenfu

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly identifies the infrastructure change: waiving 12 failed post-merge cases for the main branch.
Description check	✅ Passed	The description explains the waiver purpose, target branch, bug references, requestor, and all 12 affected test cases.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@tests/integration/test_lists/waives.txt`:
- Around line 133-148: Add the repository-standard NVIDIA copyright header at
the top of waives.txt, using the waiver-list parser’s accepted comment syntax
and setting the copyright year to 2026. Preserve all existing waiver entries
unchanged.

🪄 Autofix

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: c01298a2-c929-432a-a2d4-33fafc077a48

📥 Commits

Reviewing files that changed from the base of the PR and between f8190db and 5d1668c.

📒 Files selected for processing (1)

• tests/integration/test_lists/waives.txt

[tensorrt-cicd]

PR_Github #64547 [ skip ] triggered by Bot. Commit: 5d1668c Link to invocation

[tensorrt-cicd]

PR_Github #64547 [ skip ] completed with state SUCCESS. Commit: 5d1668c
Skipping testing for commit 5d1668c

Link to invocation

[brnguyen2]

Approving — the comments below are optional touch-ups, not blockers.

Verified the 12 entries: test IDs, params, and file paths all resolve against the tests on disk (e.g. flux1-nvfp4/flux2-nvfp4 from the FLUX_MODEL_SPECS×FLUX_FEATURE_PROFILES cross product), no duplicates, and the file stays sorted.

One thing worth a second look before merging: 10 of the 12 are visual_gen *_accuracy_against_golden cases failing in the same post-merge run under a single bug, and per that bug they are first-time failures with an identified first-bad commit — that pattern reads more like one regression than 10 independent flakes. Waiving them all removes the signal; it would be good to make sure the owner of the suspect commit is looped in so the waivers don't sit indefinitely.