Skip to content

ci: move four non-blocking PR-path rows to nightly LAB (GPU quick, CUDA unit, mac-check, sov.provenance) #4955

Closed as not planned

Description

@noahgift

Four rows on the PR / merge-queue path cannot block a merge, so under the GATE/LAB rule they leave it and run at night as LAB checks:

Row Was Nightly home
ARM64 CUDA row (gpu-quick) PR-only, not in gate's needs new gpu-lab-nightly.yml, its box concurrency group
x86 CUDA row (cuda-unit) PR-only, not in gate's needs new gpu-lab-nightly.yml, its box concurrency group
mac-check PR, merge_group, push; not in gate's needs nightly.yml, beside build-darwin, not needed by publish
sov.provenance continue-on-error, unread by sov.gate skipped on pull_request and merge_group; still runs on push to main for the attest step

Promotion to GATE was considered and rejected 3/3: none has three green nights on main and no gate retires for it.

One ticket, one owner: this issue tracks a red on any of these nightly rows. A row red or unmeasured for 7 days stops running until its owner brings it back.

Follow-ups: sov.bench (continue-on-error, gated on inputs.run_benchmarks) is to be verified as dormant on PRs.

Activity

  1. changed the title [-]ci: move four non-blocking PR-path rows to nightly LAB (gx10, yoga, mac-check, sov.provenance)[/-] [+]ci: move four non-blocking PR-path rows to nightly LAB (GPU quick, CUDA unit, mac-check, sov.provenance)[/+] on Oct 9, 2026
  2. noahgift commented on Oct 9, 2026

    @noahgift
    ContributorAuthor

    Status of the four rows:

    • GPU quick, CUDA unit, mac-check: moved to nightly in branch ci/q6-lab-rows (PR to follow).
    • sov.provenance: not moved yet. The vendored job has no if: and the vendored file is sha-pinned, and the vendored gate needs provenance in the plan, so skipping it on PRs needs a --skip-section flag in scripts/ci/fat_driver.py. That file already exceeds the pre-commit complexity limits (cyclomatic 30, cognitive 25; postfix is at cognitive 182), so the hook refuses any edit to it. Next step: bring fat_driver.py under the limits, or re-vendor with an event if: on provenance, then add the skip.
    • sov.bench: verified dormant. run_benchmarks defaults to false and ci/sections.yml never sets it, so the section is skipped on every event. No change needed.
  3. added this to the 0.71.1 milestone on Oct 10, 2026
  4. modified the milestones: 0.71.1, backlog on Oct 14, 2026
  5. noahgift commented on Oct 14, 2026

    @noahgift
    ContributorAuthor

    Folded into #5324 as a checklist row (TRIAGE-100, 2026-10-14): the open-ticket cap is 100, so this ticket is now tracked on that master and closed as not planned. Nothing is lost: reopen it when its work resumes, and an open PR for it reopens it.

    Agent: aprender-clerk

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions