Skip to content

[None][test] Add GB300 DISAGG NIXL CI Perf Test Back - #13594

Merged
chenfeiz0326 merged 8 commits into
NVIDIA:mainfrom
chenfeiz0326:chenfeiz/add-gb300-ci-perf-test-back
May 13, 2026
Merged

[None][test] Add GB300 DISAGG NIXL CI Perf Test Back#13594
chenfeiz0326 merged 8 commits into
NVIDIA:mainfrom
chenfeiz0326:chenfeiz/add-gb300-ci-perf-test-back

Conversation

@chenfeiz0326

@chenfeiz0326 chenfeiz0326 commented Apr 29, 2026

Copy link
Copy Markdown
Collaborator

Summary by CodeRabbit

  • Tests

    • Enabled multi-node performance sanity tests for GB300 configurations across 2, 3, 5, and 9 node variants
    • Activated previously disabled e2e performance sanity tests across multiple GB300 multi-node configurations
  • Bug Fixes

    • Improved environment variable handling at script initialization to reduce backend failures

Description

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

To see a list of available CI bot commands, please comment /bot help.

@chenfeiz0326
chenfeiz0326 requested a review from a team April 29, 2026 04:32
@chenfeiz0326
chenfeiz0326 requested a review from a team as a code owner April 29, 2026 04:32
@coderabbitai

coderabbitai Bot commented Apr 29, 2026

Copy link
Copy Markdown
Contributor
📝 Walkthrough

Walkthrough

This PR enables GB300 disaggregated performance sanity tests for multi-node configurations (2, 3, 5, and 9 nodes) by activating stage configurations in the Jenkins pipeline, removing UCX environment variable prefixes from pytest command generation, clearing container-injected UCX variables at SLURM script startup, and uncommenting disabled test entries across multiple test list configurations.

Changes

Cohort / File(s) Summary
Jenkins Pipeline Configuration
jenkins/L0_Test.groovy
Replaces commented-out GB300 Disagg-PerfSanity multi-node stage entries with active multiNodesSBSAConfigs += buildStageConfigs(...) calls for 2, 3, 5, and 9 node variants, updating numeric parameters and test-list identifiers.
Script Generation & Environment Setup
jenkins/scripts/perf/disaggregated/submit.py, jenkins/scripts/slurm_run.sh
Removes unset UCX_TLS UCX_NET_DEVICES && prefix from pytest command exports in submit.py and clears UCX-related environment variables (UCX_NET_DEVICES, UCX_TLS) and BASH_ENV at slurm_run.sh startup to prevent NIXL backend failures.
Test Configuration Files
tests/integration/test_lists/test-db/l0_gb300_multi_nodes_perf_sanity_*.yml
Uncomments and enables previously disabled perf/test_perf_sanity.py::test_e2e[...] integration tests across four GB300 multi-node variants (1-node, 2-node, 4-node, 8-node GPU configurations) while retaining timeout settings.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

🚥 Pre-merge checks | ✅ 3 | ❌ 2

❌ Failed checks (2 warnings)

Check name Status Explanation Resolution
Description check ⚠️ Warning The PR description is largely empty, containing only the template structure with no actual content filled in for Description, Test Coverage, or explanation of changes. Provide a clear description explaining why these tests are being re-enabled, what the UCX environment variable changes accomplish, and which tests are affected by the changes.
Docstring Coverage ⚠️ Warning Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Write docstrings for the functions missing them to satisfy the coverage threshold.
✅ Passed checks (3 passed)
Check name Status Explanation
Title check ✅ Passed The title clearly indicates the main change: re-enabling GB300 DISAGG NIXL CI performance tests by uncommenting test configurations across multiple files.
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.

✏️ Tip: You can configure your own custom pre-merge checks in the settings.

✨ Finishing Touches
🧪 Generate unit tests (beta)
  • Create PR with unit tests

Review rate limit: 9/10 reviews remaining, refill in 6 minutes.

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

@coderabbitai coderabbitai Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@jenkins/scripts/slurm_run.sh`:
- Around line 7-10: The UCX env reset is applied unconditionally; wrap the
unset/export lines in a guard so they only run for the GB300/NIXL disagg
flow—check the environment variable your pipeline uses to indicate that stage
(e.g., GB300_DISAGG, SERVING_MODE, or a stage-specific flag) and only perform
the unset UCX_NET_DEVICES UCX_TLS 2>/dev/null || true and export BASH_ENV= when
that flag indicates the GB300 disagg path; place the current lines inside an if
[ "<flag check>" ]; then ... fi block so unrelated Slurm stages keep their UCX
settings.
🪄 Autofix (Beta)

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: 6b36a7e1-390a-4b2f-995e-ea21c9dcc67a

📥 Commits

Reviewing files that changed from the base of the PR and between 0c86dda and 6e9f3c6.

📒 Files selected for processing (7)
  • jenkins/L0_Test.groovy
  • jenkins/scripts/perf/disaggregated/submit.py
  • jenkins/scripts/slurm_run.sh
  • tests/integration/test_lists/test-db/l0_gb300_multi_nodes_perf_sanity_ctx1_node1_gpu4_gen1_node1_gpu4.yml
  • tests/integration/test_lists/test-db/l0_gb300_multi_nodes_perf_sanity_ctx1_node1_gpu4_gen1_node2_gpu8.yml
  • tests/integration/test_lists/test-db/l0_gb300_multi_nodes_perf_sanity_ctx1_node1_gpu4_gen1_node4_gpu16.yml
  • tests/integration/test_lists/test-db/l0_gb300_multi_nodes_perf_sanity_ctx1_node1_gpu4_gen1_node8_gpu32.yml

Comment thread jenkins/scripts/slurm_run.sh Outdated
@chenfeiz0326
chenfeiz0326 force-pushed the chenfeiz/add-gb300-ci-perf-test-back branch from 6e9f3c6 to a469160 Compare April 29, 2026 04:40
@chenfeiz0326

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast --stage-list "GB300Disagg-PerfSanity"

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #46061 [ run ] triggered by Bot. Commit: a469160 Link to invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #46061 [ run ] completed with state SUCCESS. Commit: a469160
/LLM/main/L0_MergeRequest_PR pipeline #36205 (Partly Tested) completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

  • Please check the failed tests and fix your PR
  • If you cannot view the failures, ask the CI triggerer to share details
  • Once fixed, request an NVIDIA team member to trigger CI again

Link to invocation

@chenfeiz0326

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast --stage-list "GB300*Disagg-PerfSanity*"

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #46128 [ run ] triggered by Bot. Commit: 2591842 Link to invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #46128 [ run ] completed with state SUCCESS. Commit: 2591842
/LLM/main/L0_MergeRequest_PR pipeline #36261 (Partly Tested) completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

  • Please check the failed tests and fix your PR
  • If you cannot view the failures, ask the CI triggerer to share details
  • Once fixed, request an NVIDIA team member to trigger CI again

Link to invocation

@chenfeiz0326

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast --stage-list "GB300Disagg-PerfSanity"

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #46298 [ run ] triggered by Bot. Commit: 5cdef01 Link to invocation

@chenfeiz0326

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast --stage-list "*GB300*Disagg-PerfSanity*"

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #46314 [ run ] triggered by Bot. Commit: 5cdef01 Link to invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #46298 [ run ] completed with state ABORTED. Commit: 5cdef01

Link to invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #46314 [ run ] completed with state FAILURE. Commit: 5cdef01
/LLM/main/L0_MergeRequest_PR pipeline #36412 (Partly Tested) completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

  • Please check the failed tests and fix your PR
  • If you cannot view the failures, ask the CI triggerer to share details
  • Once fixed, request an NVIDIA team member to trigger CI again

Link to invocation

@chenfeiz0326
chenfeiz0326 force-pushed the chenfeiz/add-gb300-ci-perf-test-back branch from 5cdef01 to e4ff569 Compare May 11, 2026 08:32
@chenfeiz0326

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast --stage-list "GB300-4_GPUs-PyTorch-PerfSanity-Post-Merge-1,GB300-8_GPUs-2_Nodes-PyTorch-PerfSanity-Node2-GPU8-Post-Merge-1,GB300-8_GPUs-2_Nodes-PyTorch-Disagg-PerfSanity-CTX1-NODE1-GPU4-GEN1-NODE1-GPU4-Post-Merge-1"

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #47705 [ run ] triggered by Bot. Commit: 67151ae Link to invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #47705 [ run ] completed with state SUCCESS. Commit: 67151ae
/LLM/main/L0_MergeRequest_PR pipeline #37601 (Partly Tested) completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

  • Please check the failed tests and fix your PR
  • If you cannot view the failures, ask the CI triggerer to share details
  • Once fixed, request an NVIDIA team member to trigger CI again

Link to invocation

@chenfeiz0326

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast --stage-list "GB300-4_GPUs-PyTorch-PerfSanity-Post-Merge-1,GB300-8_GPUs-2_Nodes-PyTorch-PerfSanity-Node2-GPU8-Post-Merge-1,GB300-8_GPUs-2_Nodes-PyTorch-Disagg-PerfSanity-CTX1-NODE1-GPU4-GEN1-NODE1-GPU4-Post-Merge-1"

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #47753 [ run ] triggered by Bot. Commit: 67151ae Link to invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #47753 [ run ] completed with state SUCCESS. Commit: 67151ae
/LLM/main/L0_MergeRequest_PR pipeline #37645 (Partly Tested) completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

  • Please check the failed tests and fix your PR
  • If you cannot view the failures, ask the CI triggerer to share details
  • Once fixed, request an NVIDIA team member to trigger CI again

CI Agent Failure Analysis

Link to invocation

Signed-off-by: Chenfei Zhang <chenfeiz@nvidia.com>
@chenfeiz0326
chenfeiz0326 force-pushed the chenfeiz/add-gb300-ci-perf-test-back branch from 67151ae to 4909c22 Compare May 12, 2026 09:52
Signed-off-by: Chenfei Zhang <chenfeiz@nvidia.com>
Signed-off-by: Chenfei Zhang <chenfeiz@nvidia.com>
Signed-off-by: Chenfei Zhang <chenfeiz@nvidia.com>
Signed-off-by: Chenfei Zhang <chenfeiz@nvidia.com>
@chenfeiz0326

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast --stage-list "GB300-4_GPUs-PyTorch-PerfSanity-Post-Merge-1,GB300-8_GPUs-2_Nodes-PyTorch-PerfSanity-Node2-GPU8-Post-Merge-1,GB300-8_GPUs-2_Nodes-PyTorch-Disagg-PerfSanity-CTX1-NODE1-GPU4-GEN1-NODE1-GPU4-Post-Merge-1"

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #47953 [ run ] triggered by Bot. Commit: ae2cacb Link to invocation

Signed-off-by: Chenfei Zhang <chenfeiz@nvidia.com>
@chenfeiz0326

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast --stage-list "GB300-4_GPUs-PyTorch-PerfSanity-Post-Merge-1,GB300-8_GPUs-2_Nodes-PyTorch-PerfSanity-Node2-GPU8-Post-Merge-1,GB300-8_GPUs-2_Nodes-PyTorch-Disagg-PerfSanity-CTX1-NODE1-GPU4-GEN1-NODE1-GPU4-Post-Merge-1"

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #47956 [ run ] triggered by Bot. Commit: c2c4cf2 Link to invocation

@chenfeiz0326

Copy link
Copy Markdown
Collaborator Author

/bot run --disable-fail-fast --stage-list "GB300-*"

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #47982 [ run ] triggered by Bot. Commit: c2c4cf2 Link to invocation

Signed-off-by: Chenfei Zhang <chenfeiz@nvidia.com>
Signed-off-by: Chenfei Zhang <chenfeiz@nvidia.com>
@chenfeiz0326

Copy link
Copy Markdown
Collaborator Author

/bot skip --comment "Only add new GB300 perf tests, no need to run the whole CI pipeline"

@chenfeiz0326
chenfeiz0326 enabled auto-merge (squash) May 13, 2026 05:33
@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #48106 [ skip ] triggered by Bot. Commit: 3ee67fc Link to invocation

@tensorrt-cicd

Copy link
Copy Markdown
Collaborator

PR_Github #48106 [ skip ] completed with state SUCCESS. Commit: 3ee67fc
Skipping testing for commit 3ee67fc

Link to invocation

@chenfeiz0326
chenfeiz0326 merged commit f9ae14b into NVIDIA:main May 13, 2026
7 checks passed
yufeiwu-nv pushed a commit to yufeiwu-nv/TensorRT-LLM that referenced this pull request May 19, 2026
Signed-off-by: Chenfei Zhang <chenfeiz@nvidia.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants