Repository navigation
ci: end a work-resolve-issue kill-switch deny neutral, keeping work-unblock's re-fire #10753
Description
Problem
A work-resolve-issue run that the token broker refuses with reason kill-switch at apply-issue-patch / run or report-outcome / run still ends red. This happened in 3 runs since 2026-10-24 (for example run 40597922545). The lane breaker turns the repository kill switch on after the run has read its variables, so only the broker's live read stops it. A kill switch is a designed stop and should not count as a failure.
#10737 adds request-lane-token's kill-switch-ok input and uses it in pr-merge and intake-triage. It leaves work-resolve-issue alone. work-unblock re-fires a stopped issue by finding the failed Request the lane token step and its ##[error]lane-token-denied ... reason kill-switch line (.github/scripts/work-unblock/label-unblocked.mjs, DENIAL.token and the mints.some(... "success") short-circuit). A green step would therefore leave the issue unclaimed.
Fix
- Teach
label-unblocked.mjsto read askip-reason=kill-switchstop as a breaker trip. That means a successful mint step that minted no token, together with itslane-token-stoppednotice, or a job output naming the stop. - Then opt
pr-run-activity-write.yml's token step in, and gate its later steps and jobs as fix(pr-pipeline): end designed stops cleanly and keep held pull requests out of the merge queue #10737'scallers.test.mjscheck requires.
Done when
- A kill-switch refusal in work-resolve-issue ends green.
- work-unblock still re-fires that issue as a
breakertrip, with a test.
Related
Refs: #10737
Activity
github-actions commentedon Oct 28, 2026
Proposed work class: work-class: scoped
Basis: automated intake triage; the lanes App applied this class.
Size: Medium
Complexity: Medium
Uncertainty: Known
Flags: none
Blocked by: none
Understood: make a kill-switch refusal in work-resolve-issue runs end green, while work-unblock keeps re-firing the issue as a breaker trip. It should read the successful mint step's
lane-token-stoppednotice withskip-reasonkill-switch, and keep the old failed-step path for earlier runs.That decision settles the open questions. Nothing further is needed from you.
Automated triage, applied by the lanes App.
melodic-automation-lanes commentedon Oct 28, 2026
Not admitted to the issue lane: a criterion names .github/workflows/pr-run-activity-write.yml, which the issue lane cannot push (patch-protected-path); a builder session takes it. Admission takes the issue once each of those is met, when intake triages it again (
.github/actions/resolve-config/admission-rules.json).melodic-automation-lanes commentedon Oct 28, 2026
The latest triage verdict holds nothing: admission moved the issue to
status: confirmed; its comment says why.melodic-automation-lanes commentedon Oct 28, 2026
Refinement. A read-only run explored this repository, searched its issues and pull requests, and read the allowed docs to draft what follows. On an issue a listed human filed, it counts as the issue's spec once posted, while unedited; it is still model text, so check it.
Intended outcome: A kill-switch refusal in work-resolve-issue ends green, and work-unblock still re-fires the issue as a breaker trip, read from a skip-reason=kill-switch stop.
Dependencies
Related work
Open questions
RECOMMENDED: Yes. Hold it behind #10737 and write the criteria after that PR lands, against its real output names. Basis: .base/.github/actions/request-lane-token/action.yml:93 documents only skip-reason=workflow-modified, and a search of .base/.github finds no kill-switch-ok.
RECOMMENDED: Read the successful mint step with no token and its notice. Keep the failed-step DENIAL.token path for runs that predate the change. Basis: .base/.github/scripts/work-unblock/label-unblocked.mjs:246
Sources read
To answer, post a comment that starts with this line:
Automated refinement by the lanes App. Answer with a comment whose first line is
**Decision recorded.**.kyle-sexton commentedon Oct 30, 2026
Decision recorded. The refinement's two questions (#10753 (comment)) are settled from main at 57f3225:
request-lane-tokennow haskill-switch-ok(.github/actions/request-lane-token/action.yml:92). With it, akill-switch403 ends the step green with alane-token-stoppednotice, no token andskip-reasonkill-switch(.github/actions/request-lane-token/README.md:27,:39). fix(pr-pipeline): let a kill-switch stop skip cleanly and re-fire pr-unblock heads after the switch lifts #10844 (bd6c37d) opted pull request runs in, but kept issue runs red on purpose until work-unblock can read the new stop:kill-switch-ok: ${{ inputs.issue-number == '' && ... }}(.github/workflows/pr-run-activity-write.yml:1073). That leftover is this issue.label-unblocked.mjsread? The successful mint step'slane-token-stoppednotice withskip-reasonkill-switch. The failed-stepDENIAL.tokenpath (.github/scripts/work-unblock/label-unblocked.mjs:264-269,:1080-1088) stays for runs that predate the change.This changes no intent: a kill switch still stops the run, and only the run's colour changes.
Acceptance criteria:
.github/workflows/pr-run-activity-write.yml:1073). (body, unchanged)breakertrip, and a test covers it. (body, unchanged)lane-token-denied ... reason kill-switch, is still read as a breaker trip. (body, amended: keeps the existing path, per question 2)ci: keep a kill-switch stop from reading as a reported or passed run