Conversation
Sub test jobs now run on four downstream Jenkins instances (github-1..4) instead of a single one. BUILD_TAG carries only the job name and build number, and the instances run the same job definitions with independent build counters, so a SLURM job name no longer identifies the instance it came from -- while all four share one svc_tensorrt account and scripts directory on the cluster. Key the unique suffix on the Jenkins instance name as well, on both submission paths: - runLLMTestlistWithSbatch: the suffix flows into jobUID, the job workspace and every generated script name. No --job-name is set, so sbatch reports the basename of <jobUID>-slurm_launch.sh as the job name. - runLLMTestlistWithAgent: the suffix flows into nodeName, which reaches the cluster in the SLURM log path and the bloom/scripts and enroot-config directory names. Signed-off-by: Yue Li (TensorRT) <yuel@nvidia.com>
|
/bot run --stage-list "DGX_B200-CPP-1,DGX_H100-2_GPUs-PyTorch-GptOss-1" |
|
PR_Github #74054 [ run ] triggered by Bot. Commit: |
|
PR_Github #74054 [ run ] completed with state
|
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Enterprise Run ID: 📒 Files selected for processing (1)
Included review availability: Your plan provides up to 12 included reviews per hour; 11 remain after this review. WalkthroughThe Jenkins pipeline adds a helper that derives the Jenkins instance name from configured URLs. SLURM agent and multi-node job suffixes now include this name before the build tag and random identifier. ChangesJenkins-aware SLURM naming
Priority: ⬇️ Low Estimated code review effort: 2 (Simple) | ~10 minutes Change: Bug fix Suggested reviewers: Merge Risk: ⚪ Minimal · up to The change adds the Jenkins instance token to both relevant SLURM suffixes, and no established failure remains that should block merging. 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
Dev Engineer Review
getJenkinsInstanceName()derives the instance name fromJENKINS_URLorJenkins.instance.rootUrl, then removes trailing slashes. Both SLURM execution paths include this name in their lower-case suffix withBUILD_TAGand a six-character random identifier. No material source risks were identified.QA Engineer Review
No test changes.
Per-File QA Perspective
jenkins/L0_Test.groovy: Verify job, node, workspace, log, and generated script names across all Jenkins instances. Verify behavior when the instance name comes from either supported URL source.Description
Sub test jobs now run on four downstream Jenkins instances (github-1..4) instead of a single one. This adds the Jenkins instance name to the unique suffix on both SLURM submission paths:
runLLMTestlistWithSbatchrunLLMTestlistWithAgentTest Coverage
SLURM job name:

SLURM output log file name:

The UNSTABLE is just the multi-GPU ci: full pre-merge approved label gate, not a test failure.
PR Checklist