[antonwolfy]

#3152 added a per-test timeout to the tensor tests of the conda package workflows, for the hang at 99% (IntelPython/dpctl#1954). check-onemath.yaml hits the same hang but was left out, because it runs the tests without xdist.

Measured on check-onemath.yaml runs since 2026-10-01, from the retry annotations: 55 of 577 test jobs hit the 10 min timeout of the tensor step, 8 of them on two attempts, so about 10.5 h of runner time went to waiting. A healthy tensor step takes a median of 291 s (37 recent jobs).

This PR:

• runs the tensor tests with --timeout=120 --timeout-method=thread, as conda-package.yml does;
• adds -n auto to them: without xdist the thread method exits the whole pytest process at the first timeout, so the run stops there and the retry starts from scratch. With xdist only the stuck worker is killed and replaced, the run finishes, and the hung test is reported by name;
• adds pytest-timeout and pytest-xdist to environments/test.yml. generate_coverage.yaml also merges that file but passes neither option, so its run doesn't change.

Checked locally with a test that hangs: a serial run with the thread method ends at the hang, the tests after it don't run and no cache is written, while with -n 2 the worker is replaced and the other tests all run.

Expected effect: a hang fails one test after 2 min instead of stalling the step for 10 min, and xdist should also shorten the tensor step; this PR's CI shows by how much.

• Have you provided a meaningful PR description?
• Have you added a test, reproducer or referred to an issue with a reproducer?
• Have you tested your changes locally for CPU and GPU devices?
• Have you made sure that new changes do not introduce compiler warnings?
• Have you checked performance impact of proposed changes?
• Have you added documentation for your changes, if necessary?
• Have you added your changes to the changelog?

[github-actions]

Array API standard conformance tests for dpnp=0.22.0dev1=np2py314h8d9cdd5_14 ran successfully.
Passed: 1491
Failed: 0
Skipped: 6

[github-actions]

View rendered docs @ https://intelpython.github.io/dpnp/index.html

[Copilot]

🟢 Approval recommended

No active review issues remain for the oneMath tensor-test timeout and parallel-execution changes.

No open issues 🎉

What changed in this PR

Extends the existing tensor-test hang mitigation to the oneMath workflow, using per-test timeouts and parallel workers so a stuck test can be terminated without stopping the entire test run.

Changes:

• Applies a 120-second thread-based timeout to all four tensor-test commands, including retry paths.
• Enables pytest-xdist with -n auto so timed-out workers can be replaced.
• Adds the required pytest plugins to the shared test environment.

File	Description
environments/​test.yml	Adds pytest-timeout and pytest-xdist dependencies.
.github/​workflows/​check-onemath.yaml	Configures and applies per-test timeouts and parallel execution for tensor tests.

_{Review effort: Balanced}

[coveralls]

coverage: 84.397%. remained the same — onemath-tensor-test-timeout into master

[vlad-perevezentsev]

LGTM
Thank you @antonwolfy