[https://nvbugs/6713424][chore] Unwaive GB300 DSpark AgentX con1156 e2e

[chenfeiz0326]

Description

Unwaives the GB300 DeepSeek-V4-Pro DSpark AgentX con1156 e2e perf-sanity case:

perf/test_perf_sanity.py::test_e2e[disagg_upload-e2e-gb300_deepseek-v4-pro-dspark_agentx_con1156_ctx2_dep8_gen1_dep8_eplb0_dspark3_ccb-NIXL]

It was waived after main post-merge 3003 under nvbugs/6877254. That bug was closed with a pointer to nvbugs/6713424, the AgentX empty-response failure, which #20424 fixed (merged as 2f5bb04):

• The HF streaming detokenizer now releases held partial UTF-8 characters on the final response. A one-token response no longer streams empty text, which aiperf rejected as InvalidInferenceResultError and which failed the lane on a single request.
• The AgentX client sends skip_special_tokens:false, so an EOS sampled under ignore_eos is visible text.
• The client recognizes the upstream PyPI aiperf from requirements-dev.txt and installs the agentx fork instead of failing on Unknown option: --scenario.

The lane runs in the GB300-24_GPUs-6_Nodes-PyTorch-Disagg-PerfSanity-AgentX-CTX2-NODE2-GPU8-GEN1-NODE2-GPU8-Post-Merge-1 stage (l0_gb300_multi_nodes_perf_sanity_ctx2_node2_gpu8_gen1_node2_gpu8), which a regular pre-merge run does not include.

Test Coverage

• GPU verification of the fix on GB300 (oci-jhb), DSv4-Pro DSpark AgentX: 9 of 6,015 one-token responses were empty without the fix and 0 of 5,979 with it. With both changes the run had 0 empty responses of 2,973 and passed with no aiperf errors. Details are on nvbugs/6713424.
• This PR: CI runs the unwaived case in its own post-merge stage via --stage-list.

PR Checklist

• PR description clearly explains what and why.
• PR follows TRT-LLM coding guidelines.
• Test cases are provided for new code paths.

🤖 Generated with Claude Code

[chenfeiz0326]

/bot run --stage-list "GB300-24_GPUs-6_Nodes-PyTorch-Disagg-PerfSanity-AgentX-CTX2-NODE2-GPU8-GEN1-NODE2-GPU8-Post-Merge-1"

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

• Configuration used: Repository: NVIDIA/TensorRT-LLM/.coderabbit.yaml
• Review profile: CHILL
• Plan: Enterprise
• Run ID: 803430ff-1304-4e59-9249-880078e97069

📥 Commits

Reviewing files that changed from the base of the PR and between 0372799 and 85d4827.

📒 Files selected for processing: 1

• tests/integration/test_lists/waives.txt

💤 Files with no reviewable changes: 1

• tests/integration/test_lists/waives.txt

Included review availability: This review used your included allowance. Your plan provides up to 12 included reviews per hour; 11 remain after this review.

⚠️ A high-level summary could not be generated for this review. CodeRabbit will regenerate it on the next update, or you can request a refresh with @coderabbitai summary.

Walkthrough

The performance test waiver list no longer skips the specified disaggregated upload scenario. The entry included a reference to NVBUG 6877254.

Changes

Performance test waiver update

Layer / File(s)	Summary
Remove performance test skip tests/integration/test_lists/waives.txt	Removed the skip entry for the specified disaggregated upload performance sanity test and its NVBUG 6877254 reference.

Priority: ⬇️ Low

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

Change: Other

Suggested reviewers: trtllm-agent

Merge Risk: ⚪ Minimal · up to 85d48

This re-enables a previously waived GB300 DSpark AgentX performance sanity test. It has no production impact, and the only remaining risk is that the post-merge lane fails, which CI would show.

Pre-merge checks: 5

Check name	Status	Explanation
Title check		The title clearly identifies the main change: re-enabling the GB300 DSpark AgentX con1156 end-to-end test. It uses the required NVBugs and chore format.
Description check		The description explains the issue, the fix, the affected test, supporting validation results, and CI coverage. The checklist is only partially completed, but the required technical context and test c…
Docstring Coverage		No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check. Docstring coverage is scoped to functions touched by this diff. Analyzed 0 functions across 0…
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.

Full details: Description check

Explanation

The description explains the issue, the fix, the affected test, supporting validation results, and CI coverage. The checklist is only partially completed, but the required technical context and test coverage are present.

Full details: Docstring Coverage

Explanation

No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check. Docstring coverage is scoped to functions touched by this diff. Analyzed 0 functions across 0 files.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

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

[tensorrt-cicd]

PR_Github #78823 [ run ] triggered by Bot. Commit: 85d4827 Link to invocation

[github-actions]

Automatically added "ci: full pre-merge approved" because this PR has satisfied the required GitHub review approvals. Unresolved review conversations and other required checks remain independent merge requirements.

[chenfeiz0326]

/bot skip --comment "Only unwaive perf tests, no need to run the whole CI pipeline"

[tensorrt-cicd]

PR_Github #78896 [ skip ] triggered by Bot. Commit: 85d4827 Link to invocation

[tensorrt-cicd]

PR_Github #78823 [ run ] completed with state ABORTED. Commit: 85d4827

Link to invocation

[tensorrt-cicd]

PR_Github #78896 [ skip ] completed with state SUCCESS. Commit: 85d4827
Skipping testing for commit 85d4827

Link to invocation