[cortinico]

Summary:

The scheduled Monitor React Native New Issues workflow fails in Extract next oncall because extractIssueOncalls.js requires an exact dated entry and the ONCALL_SCHEDULE secret has no entry for the active Tuesday.

Live evidence:

• Affected default-branch head: 99d327c613cdd4c3614cc3baa286a2d57a629136
• Current failed run: https://github.com/react/react-native/actions/runs/37395939025
• Error: TypeError: Cannot read properties of undefined (reading '0') at schedule[dateStr][0]
• Last successful scheduled run: https://github.com/react/react-native/actions/runs/20724738462 on 2026-01-05
• First failed scheduled run: https://github.com/react/react-native/actions/runs/20733701571 on 2026-01-06
• No matching open PR was found before preparing this change.

The exact-date lookup was introduced in 890b8f7ea3378685f7d6d9612ff584fc9ed17df0. The outage itself was triggered by the schedule crossing its final configured date, not by a new January commit, so a code bisection cannot isolate a later introducing commit.

This change treats each dated rotation as effective until the next dated entry, which keeps issue monitoring alive when schedule maintenance is late. It writes a warning to stderr whenever it carries a rotation forward, so the stale configuration remains visible. Dates before the first configured entry still fail clearly.

Risk / limitation: until the secret is refreshed, the last configured pair remains on call. That is preferable to the current complete notification outage, but schedule owners should still update the secret.

Changelog:

[INTERNAL] [FIXED] - Keep issue monitoring running when the on-call schedule reaches its final dated entry.

Test Plan:

• npx --yes jest@29.7.0 --runInBand --config '{"testEnvironment":"node","rootDir":"/tmp/react-native-ci-monitor/repo","transform":{}}' .github/workflow-scripts/__tests__/extractIssueOncalls-test.js — 9/9 tests passed, including the new expired-schedule regression case.
• npx --yes prettier@3.9.4 --check .github/workflow-scripts/extractIssueOncalls.js .github/workflow-scripts/__tests__/extractIssueOncalls-test.js — passed.
• node --check on both changed files — passed.
• git diff --check — passed.
• Monitor React Native New Issues workflow-dispatch run on this branch — passed end to end: https://github.com/react/react-native/actions/runs/37396984698. The real secret took the carry-forward path, emitted the expected warning, and completed Monitor New Issues successfully.

A full workspace install was not run locally because the sparse diagnostic checkout cannot resolve unpublished 0.87.0-main workspace packages from npm; GitHub PR CI will exercise the repository-wide checks.

2026-10-18 CI monitor update

Revalidated against current main SHA 27e77216ef2557ccbf31a01067a44dc1dd2ec846.

• Complete exact-SHA inventory: 421 terminal checks (278 success, 137 skipped, 6 failure); no queued, running, cancelled, or timed-out checks.
• All six failures are scheduled Monitor React Native New Issues runs. Their logs have the same deterministic TypeError: Cannot read properties of undefined (reading '0') at extractIssueOncalls.js:23, including the latest run: https://github.com/react/react-native/actions/runs/39126927293
• No other open PR matches this root cause.
• The patch applies cleanly to the current default-branch head.
• Fresh validation: focused Jest suite 9/9 passed; Prettier, node --check on both changed files, and git diff --check passed.
• Direct reproduction with an exhausted schedule fails on current main and succeeds with this patch by carrying forward the final configured rotation while emitting a warning.

The introducing code remains commit 890b8f7ea3378685f7d6d9612ff584fc9ed17df0; the outage starts only when the finite secret-backed schedule crosses its last configured date, so there is no later code commit to bisect.

[cortinico]

Superseded by #59499. We are retiring the new-issue Discord monitor entirely, so its on-call schedule no longer needs a fix.