hourly full run: red on main (CI)

[github-actions]

os-hourly-full-run-ci — machine-findable marker for this generated card. ⛔ Do not delete this line: it is how the hourly full run finds this card instead of filing a new one every hour.

hourly full run: red on main (CI)

Swept 2026-09-09T05:15:19.205Z · run log · commit 041d9fdc689043705ccc460f18ad7cb65a3fa3fa · conclusion failure.
CI runs the FULL battery on main every hour (#16467). A push run on
main tests only the packages the merge touched, and a merge-queue build tests only what the group
contained — so this hourly run is the only thing that says whether the WHOLE tree is green, and
this card is the only channel that reports it. Nothing is blocked by it.
⛔ The remedy is never to narrow, skip or delete the failing check to make the hourly run green.
Fix what it names and let the next hourly run refresh this card; a card nobody reopens is closed by
the next green run being uneventful, not by editing this one.

What landed since the last green hourly run

Range: 881ce2bf31 (run 34309339980, the last green CI schedule run) .. 041d9fdc68 — compare

• 7392a1afb7 docs(skills): the walk-radius mandate picks the turbo task by rule, not by name (docs(skills): the walk-radius mandate picks the turbo task by rule, not by name #17016)
• 3c27a43e41 skills(pm-dispatch): the queue-entry rule names its third case — red BY DESIGN on a non-required check — and the two ceilings rise by one as ruled (skills(pm-dispatch): the queue-entry rule names its third case — red BY DESIGN on a non-required check — and the two ceilings rise by one as ruled #17021)
• c0e4bc749d docs(agents): the commit-trailer exemption is a reporting one; the acceptance-notes heading is English at all five sites (docs(agents): the commit-trailer exemption is a reporting one; the acceptance-notes heading is English at all five sites #17023)
• ca31ff66ed fix(deps): take the fix for the fifteen OSV advisories blocking every PR (fix(deps): take the fix for the fifteen OSV advisories blocking every PR #17029)
• 74d1340f7a docs(pm-dispatch): state both the batch default and the maintainer ceiling in one row (docs(pm-dispatch): state both the batch default and the maintainer ceiling in one row #17024)
• 513c495560 fix(turbo): stop the root-scoped test-input radius hashing node_modules (fix(turbo): stop the root-scoped test-input radius hashing node_modules, and gate the class #17035)
• cc986c9134 test(cli, create-objectstack): give MONOREPO_ONLY a pattern for project-escaping relative paths, and re-equalise the two pins' vocabularies (test(cli, create-objectstack): give MONOREPO_ONLY a pattern for project-escaping relative paths, and re-equalise the two pins' vocabularies #17018)
• 041d9fdc68 fix(analytics): ask the object-level read grant before serving an inline dataset — one admission verdict on every driver (fix(analytics): ask the object-level read grant before serving an inline dataset — one admission verdict on every driver #16860)
  Filed by .github/workflows/scheduled-full-run-card.yml. Generated by Claude Code

[baozhoutao]

Answered: one flaky test, diagnosed and carded. Closing.

domain:devx PM seat (#6023), session session_012GKcPZbMoGq7WPzKLfRBTU.

What was red

Exactly two jobs of 17 in run 34313211319 — Test Core (4/6) and the Test Core aggregate it fails. Everything else concluded success (tally: 15 success, 2 failure). One package, one test, one assertion:

FAIL src/turso-driver-timeout.test.ts > TursoDriverConfig.timeout — remote mode over HTTP
     > a stalled remote fails the operation within the configured window, as TIMEOUT / 504
AssertionError: expected 0 to be greater than 0   ❯ :122:31
Test Files  1 failed | 49 passed (50)
Tests       1 failed | 1230 passed (1231)

The operation did fail as TIMEOUT / 504 inside its window — every assertion about the driver passed. What failed is the epilogue expect(remote.requests()).toBeGreaterThan(0): the test gives the client 100 ms (WINDOW_MS) to get bytes onto a local socket while six shards run concurrently on the runner.

It is non-deterministic, and that is measured rather than assumed

next hourly run, 34317273493 at 06:01Z	success
⭐ git diff --stat 041d9fdc68 a016f08b8a -- packages/drivers/driver-turso/	empty — untouched
control: what did change between those commits	24 files, 2,966 insertions

⇒ Identical driver code, identical test, red then green. ⛔ Nothing to bisect and no infrastructure fault to blame.

⚠️ I nearly closed this on "the next run was green", which would have been wrong: the green run is on a different commit, so on its own it proves only that a016f08b8a is green. The untouched-package diff is what makes it a flake.

Filed

#17079 — the race, both call sites (:122 and :138 carry the same assertion; only the first has been observed losing), and three candidate directions left explicitly unprejudged. ⛔ Ungraded and unlabelled: domain:* is triage's.

Closing, and a note on this card's own machinery

Closing per this card's rule — "a card nobody reopens is closed by the next green run being uneventful" — with the diagnosis carded so nothing is lost.

⚠️ Worth stating for the next reader, because I checked and briefly misread it: scheduled-full-run-card.yml has no close path and is not supposed to have one. Its job guard is conclusion == 'failure' || conclusion == 'timed_out', so it only ever files or refreshes; closing is a seat's act, which is consistent with its own ⛔ "A CLOSED card is never reopened. Red again after it was answered is a regression, filed fresh." The six most recent runs of that workflow all conclude skipped — that is the guard working on non-schedule events, ⛔ not a broken watcher.

⇒ The #16467 machinery did exactly what it was built for: a red on main that blocks nothing and notifies nobody got a card, a diagnosis and a follow-up card within two hours.

---

Generated by Claude Code