[https://nvbugs/6541322][fix] Remove stale waiver for TestQwen3NextInstruct::test_bf16_4gpu[dep4]

[trtllm-agent]

Summary

• Root cause: The Qwen3-Next-80B bf16 TP4/EP4 test aborted during LLM startup on the V2 KV-cache Mamba SSM slot check, which computed only 75 available SSM slots against the 129 required by the configured batch size, raising before any model weights were loaded. The underlying defect was the V2 SSM pool being rounded below the minimum-slot floor implied by max_num_requests; that floor fix has since landed in HEAD via commit 7f7dccf, so the failing code path no longer exists. The waiver entry for this test therefore outlived the defect it was tracking.
• Fix: Verification on current HEAD confirmed the reported failure signature is gone — the run cleared init completely (weights loaded, 19 CUDA-graph sizes captured, KV cache sized, 3786 MMLU requests submitted), which is structurally incompatible with the reported abort that produced zero model-load lines. The only remaining crash was unrelated GPU co-tenancy (a peer process holding ~121 GB/GPU, visible only to in-container nvidia-smi, leaving 53 MB free at a cublas allocation), so no product change was warranted. The change is waiver-only: drop the dep4 SKIP line so the test runs again in CI, while leaving the sibling tp1_block_reuse-cutlass waiver under the same bug in place since it was not shown to be fixed.
• Automated fix generated by repair-bot

Test plan

• Verify fix on the same GPU type as the original failure
• Check for regressions in related tests

Links

• Bug: https://nvbugs/6541322

Dev Engineer Review

• Removed the stale dep4 waiver for TestQwen3NextInstruct::test_bf16_4gpu[dep4].
• The waiver removal matches the fix for the V2 Mamba SSM slot-sizing defect.
• The sibling tp1_block_reuse-cutlass waiver remains unchanged.
• The test-list change is limited to one valid waiver entry.

QA Engineer Review

• Modified tests/integration/test_lists/waives.txt.
• Removed one waiver entry.
• No test code changed.
• CBTS coverage data is unavailable. Verdict: needs follow-up.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 652d6793-daad-475d-9837-5674bd3488b2

📥 Commits

Reviewing files that changed from the base of the PR and between 946f9c5 and 672b815.

📒 Files selected for processing (1)

• tests/integration/test_lists/waives.txt

💤 Files with no reviewable changes (1)

• tests/integration/test_lists/waives.txt

---

Walkthrough

The change removes the skip entry for TestQwen3NextInstruct::test_bf16_4gpu[dep4], associated with NVBug 6535767, from the integration waiver list.

Changes

Cohort / File(s)	Summary
Test waiver removal tests/integration/test_lists/waives.txt	Removes the waiver for the Qwen3Next BF16 four-GPU accuracy test.

Estimated code review effort: 1 (Trivial) | ~2 minutes

Possibly related PRs

• NVIDIA/TensorRT-LLM#17131: Removes the same test waiver.
• NVIDIA/TensorRT-LLM#17041: Modifies the same waiver entry for NVBug 6535767.
• NVIDIA/TensorRT-LLM#16924: Removes a waiver for the same Qwen3Next BF16 four-GPU test with a different configuration.

Suggested reviewers: bowenfu, yingguo-trt, mzweilz

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly identifies the NVBugs issue, fix type, stale waiver, and affected test.
Description check	✅ Passed	The description explains the root cause and waiver-only fix, and includes relevant verification details and the bug link.
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.}

[xinhe-nv]

/bot run --only-qa-verify test accuracy/test_llm_api_pytorch.py::TestQwen3NextInstruct::test_bf16_4gpu[dep4]

[tensorrt-cicd]

PR_Github #63612 [ run ] triggered by Bot. Commit: 672b815 Link to invocation

[trtllm-agent]

NVBug 6541322 was closed as Bug - Fixed. The repair-bot root cause matches the merged resolution #17131, so this PR is being auto-closed. If you believe this is wrong, reopen the PR — the bot will never auto-close a human-reopened PR for this cleanup generation.

[tensorrt-cicd]

PR_Github #63612 [ run ] completed with state ABORTED. Commit: 672b815

Link to invocation