Conversation
The mpi_ray_parity marker on this test adds a separate ray_mode parametrization when CI runs with --run-ray, producing a distinct test id (test_nemotron_h_breakable_prefill_cuda_graph[ray-tp1]) that the existing waive entry for [tp1] (https://nvbugs/6777501) does not cover, since waive matching is an exact nodeid match with no wildcard support. Add the matching [ray-tp1] entry so the Ray-executor variant of this known failure is waived too. Signed-off-by: farazkh80 <58580514+farazkh80@users.noreply.github.com>
|
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 skipped test list adds the Ray/Tensor-Parallel-1 Nemotron-H breakable-prefill CUDA-graph test, tracked by NV bug 6777501. ChangesNemotron-H test waiver
Priority: ⬇️ Low Estimated code review effort: 1 (Trivial) | ~2 minutes Change: Other Suggested reviewers: Merge Risk: ⚪ Minimal · up to This narrowly scoped test waiver is mergeable with no outstanding actionable risk identified. 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
Description
test_nemotron_h_breakable_prefill_cuda_graphis marked@pytest.mark.mpi_ray_parity, which adds a separateray_modeparametrization when CI runs with--run-ray(tests/unittest/conftest.py), producing a distinct test id[ray-tp1]alongside the base[tp1]. Waive matching inapply_waives()(tests/integration/defs/test_list_parser.py) is an exact nodeid match with no wildcard support, so the existing waive for[tp1](nvbugs/6777501) does not cover[ray-tp1]. This is currently failing theH100_PCIe-PyTorch-Ray-1CI stage on PR #19038 as an unrelated pre-existing issue.Changes
[ray-tp1], same nvbug.Open question
I have not confirmed that nvbugs/6777501 covers the Ray-executor code path specifically, only that the assertion text is identical to the MPI variant. If the Ray path turns out to have a distinct root cause, this waive should be revisited.
PR Checklist
pre-commit run --all-filesand fixed any issues.Dev Engineer Review
Adds an exact waiver for the Ray-specific test ID under NV bug
6777501. The existing[tp1]waiver remains separate. Scope is limited to the reported CI failure.QA Engineer Review
Modifies only
tests/integration/test_lists/waives.txt. It adds theray-tp1waiver; the existingtp1waiver remains present. No test execution results are supplied. Coverage verdict: needs follow-up.Per-File QA Perspective
tests/integration/test_lists/waives.txt: Verify that the exactray-tp1entry suppresses only the intended test and does not affect other parametrizations. No separate CI or manual-QA list change applies to this waiver-only update.