Repository navigation
fix(fleet-ci-fail-alert): queue dedupe keys every red job and survives deleted queue refs - #1430
Conversation
…s deleted queue refs FleetReview #1375 @fbffd2d7 (t_8f641596), both P1s real-open on main: - afff7c56: sig_of() dropped failed jobs without a "FAILED (not quarantined)" annotation whenever any job had one, so a new lint red next to an already-paged test failure matched the earlier signature and was silenced. Each failed job now contributes its tests, or its slice-stripped name. - f6fcb4be: prior runs whose queue ref was deleted come back head_branch null (live: run 36354859649) and were discarded. The PR is now taken from the squash subject's trailing "(#N)" when head_branch is empty, for prior runs and the current run. Branch and subject agreed on 100/100 recent merge_group CI runs. Verified: tests/test_fleet_ci_fail_alert_gate.py 34 passed; the 3 new tests fail on the base workflow. Real route step replayed read-only against live runs 36387356925 / 36384622295 (PR #1401 retries): still no page, same signature as before.
FleetReviewBelow the merit threshold, review skipped (90 changed lines < 150 and 2 changed files < 6, no hot path touched; skipped by policy). FleetReview · reviewKind: skipped-by-policy |
FleetReviewBelow the merit threshold, review skipped (90 changed lines < 150 and 2 changed files < 6, no hot path touched; skipped by policy). FleetReview · reviewKind: skipped-by-policy |
ang-fleet-interactive
Bot
added
the
ci-reviewed
FleetReviewBelow the merit threshold, review skipped (90 changed lines < 150 and 2 changed files < 6, no hot path touched; skipped by policy). FleetReview · reviewKind: skipped-by-policy |
ang-fleet-ci-actuators
Bot
added
the
fleetreview:post-merge
FleetReviewReview: post-merge · head
Post-merge review ( profile: light (rule: default light: lines 90<800, files 2<1000000, hunks 5<1000000, no hot path) · round 0 · members: B-assert-ctx, B-state, L6, C-assert-xhigh · families: anthropic,openai Confidence: 1/5 Findings
FleetReview provenance · models: B=gpt-6-sol, C=claude-code-opus-5-5 · cost: $0.93 · duration: 2m 28s · rounds: 1 · files examined: 2 |
FleetReview follow-up on #1375 @fbffd2d7 (card t_8f641596). Both P1 findings were still open on main.
FAILED (not quarantined)annotation,sig_of()left out failed jobs that had none. So a lint job going red next to an already-paged test failure matched the earlier signature and got suppressed. Now every failed job adds either its annotated tests or its name with the slice number stripped (Tests completeandAll required checksare still skipped).gh-readonly-queueref was deleted comes back withhead_branch: null(seen live on run 36354859649), and the lookup threw it away. Whenhead_branchis empty, the PR number now comes from the squash commit subject's trailing(#N). This applies to prior runs and to the current run. The branch and the subject agreed on 100/100 recent merge_group CI runs.Verification
tests/test_fleet_ci_fail_alert_gate.py: 34 passed. The 3 new tests fail against the base workflow.route=nonewith the same test signature.failing job unknown, which is intended.