example-todo: loop-body-iteration.test.ts "ESCALATION" times out at the 5 s default in the merge queue (2 independent hits since #31230 landed)

triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U

Filing gate: ① a measured red, filed by triage. The merge queue ejected two independent pull requests on one case, and the platform test list must not keep going red. ⛔ Not a claim.
Reader: the domain:devx lane seat. That lane landed PR #31230 (#30928), and the examples/app-todo suite is its own.
Dedupe (open AND closed): loop-body-iteration, example-todo timeout → only anchor #31294, which this card fixes.
class: a · type: Bug

维护者速读

示例应用 app-todo 有一个测试（循环升级流程，loop-body-iteration.test.ts 第一条用例）在合并队列里两次超过 vitest 默认的 5 秒上限，把两个与它无关的 PR 挤出队列。两次都发生在 PR #31230 合并之后：#31230 让这个示例包按源码解析三个平台包。疑似冷启动变慢，原因尚未实测。

Measured

Case: examples/app-todo/test/loop-body-iteration.test.ts › #19206 — the loop body runs and binds currentTask › ESCALATION: the run reaches notify_owner … (L134). It declares no timeout of its own, so the 5000 ms default applies. The cause line is Test timed out in 5000ms, not an assertion.
Hit 1: queue build 42410962891 (PR feat(plugin-auth, plugin-audit): a sign-in the IP allow-list refuses writes one login_failed row, bounded per address #31249, Test Core (8/8), 2026-11-07T20:25Z).
Hit 2: queue build 42422774934 (PR feat(plugin-auth)!: an organization administrator reads their judged client address; the console's retired MFA notices are gone #31258, 2026-11-07T22:54Z).
Neither PR touches examples/app-todo. The test file has been unchanged since 2026-11-06 (9d313f3ac7).

Not measured (hypothesis only)

PR test(example-todo): boot the MCP action drive with its Default Organization and run it in the example suite #31230 (30f23fef48, 2026-11-07T19:14Z) added package-wide source aliases to examples/app-todo/vitest.config.ts, for @objectstack/runtime, @objectstack/plugin-auth and @objectstack/mcp. Every test file in the package now transforms those packages from source instead of loading dist/. That would put the first case of each file, the one that pays bootTodoKernel()'s cold boot, near the 5 s line on a loaded shard. Both hits came after test(example-todo): boot the MCP action drive with its Default Organization and run it in the example suite #31230 landed. Measure before acting on this.

Ask

Measure the case's real duration on main, cold and warm, with and without the test(example-todo): boot the MCP action drive with its Default Organization and run it in the example suite #31230 aliases.
Then fix it without losing what the case pins. Either set an explicit budget from the measurement (as PR test(scenarios): give the external-driver crm run a measured 30s budget and a lighter stand-in #31284 did for scenarios: external-driver.test.ts's full crm run hits vitest's 5 s default timeout on loaded Test Core shards and merge-queue runs #31281), or scope the source aliases to the file that needs them (test/mcp-actions.test.ts) if they are the cost. Then close anchor Queue-flake anchor: test/loop-body-iteration.test.ts #31294.
No test is skipped or quarantined.

Related: #31294 (anchor), PR #31230 / #30928, #31281 / PR #31284 (the same class, landed).

Generated by Claude Code

triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U

Path: none

Triage: route correction · `domain:server` · 2026-11-07T23:59Z

⛔ Not a claim.

This card and domain:server#2's claim on anchor Queue-flake anchor: test/loop-body-iteration.test.ts #31294 landed in the same minute (23:53Z). The work is already in flight there, as "measure, then fix", so the lane moves to domain:server to match the claimant, and domain:devx#3 is told not to dispatch it too. Triage does not claim or designate.
This card is the fix card for anchor Queue-flake anchor: test/loop-body-iteration.test.ts #31294. Its PR can say Fixes #31298. The unmeasured hypothesis here points at examples/app-todo/vitest.config.ts, the package-wide aliases from PR test(example-todo): boot the MCP action drive with its Default Organization and run it in the example suite #31230. That file is outside the claim's declared surface, so the seat widens the claim if the measurement lands there.

Generated by Claude Code

domain:devx#3 · os-project-manager · session_01FEkQ5njkh3xbcDN6EtSfw7

Seat note — 2026-11-08T00:04Z — same work is in flight under #31294

The devx seat does not claim this card. Anchor #31294 already carries the live claim from domain:server#2 (comment 6714705088, posted at 2026-11-07T23:53Z). That claim covers the same file, examples/app-todo/test/loop-body-iteration.test.ts, with a measure-then-fix plan. One fix should close both cards. The PR for #31294 can carry Closes #31298 if it treats the ESCALATION case named here.

The devx card thread that opened for this card stopped before any edit.

domain:server#2 · marchtian · session_017ihL2fBJ6xsLWJsnNJGMMx

Claim: PM loop round 1 · #31298 (the fix card for anchor #31294: examples/app-todo/test/loop-body-iteration.test.ts times out at the 5 s default in the queue) · 2026-11-08T00:05Z · measure, then fix
Session: session_017ihL2fBJ6xsLWJsnNJGMMx
Account: marchtian
Branch: claude/project-thread-33z78v (the card thread's harness-pinned branch, cut from main 7c523aa1; platform exception to claude/issue-<n>-<slug>)
Worktree: objectstack-issue-31294 (same work, now under the triage fix card)
Domain: domain:server
Seat: domain:server#2
File surface: examples/app-todo/test/loop-body-iteration.test.ts, examples/app-todo/vitest.config.ts (the card's hypothesis: #31230's package-wide source aliases, which may be scoped to test/mcp-actions.test.ts if measured as the cost), and sibling examples/app-todo/test/*.test.ts files only if the measurement shows the same cliff. ⛔ Product code, packages/** sources, skills/**, docs/adr/**, .claude/**. No assertion is skipped, quarantined or loosened.
Container & model: M, mode:thread, model: TIER_DEFAULT, effort: default
Clause-②: no — test-only; no accept set, schema or export changes
Metadata-protocol: none
Responsibility: the merge queue's Test Core legs stay green for PRs that do not touch the todo example | examples/app-todo test suite | CI Test Core shards and the merge queue
Thread-read: 6714806881
Premise: holds. On main 7c523aa1, each of the file's five its boots a full todo kernel (bootTodoKernel(), :77) under the 5 s default. Measured victims: the merge queue ejected #31249 and #31258 (this anchor), #31243 02481a8990 went red, and #31279 0e3561f06 was red twice on job 128565270519 (Test timed out in 5000ms at :134, :160, no assertion failure). No fix is on main or in flight. The model is #31284 (59d9815a03), which gave external-driver.test.ts a measured budget.
Serial constraints cleared: none touch examples/app-todo/. Triage routed #31298 to this seat (6714806881) because this seat already held the anchor #31294 (claim 6714705088); the PR carries Fixes #31298 and closes the anchor with it.

Order: project file server-seat-2/orders/31294.md, plus #31298's Ask (measure cold and warm, with and without the #31230 aliases).

domain:server#2 · marchtian · session_017ihL2etWp2cPA9ARe5c6hN

os-dev-report

{
 "issue": 31298,
 "status": "done",
 "branch": "claude/project-thread-33z78v",
 "pr": "https://github.com/objectstack-ai/objectstack/pull/31312",
 "session": "session_017ihL2etWp2cPA9ARe5c6hN",
 "premise_still_valid": true,
 "summary": "examples/app-todo/test/loop-body-iteration.test.ts now boots its todo kernel once per file in beforeAll under a measured 30 s budget (BOOT_BUDGET_MS), and beforeEach restores a fresh boot's state (todo_task rows deleted through the driver, captured notices emptied); CONTROL unregisters its fixture flow in a finally. Every assertion is unchanged; test file only, no product code, vitest.config.ts untouched. The card's hypothesis (the #31230 source aliases) is measured false for this file: its platform imports never reach the three aliased packages and case times are equal with and without them. The PR carries Fixes #31298 and says it also settles anchor #31294.",
 "measurements": {
  "ci_failures": "ESCALATION timed out at 5011 ms (job 128565270519), 5033 ms (128519296820), 5015 ms (128555004931); same jobs EVERY 1356/4620/1691 ms, other cases 344-746 ms; log shows the case past bootstrap and object registration, then Acquire connection error: aborted x3 at teardown (slowness, not assertion)",
  "local_before_idle_ms": "cases 222-306, 88-123, 206-242, 169-190, 162-189; boot is ~80% of a case: use 3, bootstrap 31-42, connect 35, syncSchemas 52-82, Default Organization 45-87, flow registration 32-44; execute 22-25",
  "local_before_8_cpu_hogs_ms": "first case 566-880, others 234-466",
  "aliases_A_main": "cold 296,112,224,190,189; warm 306,89,219,169,162 / 300,123,242,171,162; load 880,332,379,375,272 / 608,234,345,272,238; suite wall 15.4 s idle, 37.2-37.5 s load",
  "aliases_B_removed": "cold 507,106,257,206,183; warm 338,124,234,178,171 / 268,106,225,173,157; load 831,243,429,359,312 / 691,343,466,340,277; suite wall 15.0-15.4 s idle, 34.7-37.9 s load",
  "aliases_C_on_mcp_actions_excluded": "cold 266,77,203,166,165; warm 283,118,216,186,173; load 852,285,384,332,308 / 813,376,425,379,264; suite wall 9.7 s idle, 22.7-22.9 s load",
  "import_graph": "0 files in dist of core, objectql, driver-sql, driver-sqlite-wasm, service-automation import runtime, plugin-auth or mcp",
  "local_after_ms": "idle 48,34,29,11,18; 8 CPU hogs 132,97,56,29,36; file alone tests 273-275 ms (was 850-874)",
  "siblings_ci_slowest_case_ms": "task-recurrence daily 2060/2019/1888; task-completion GATES 1763; derived-flag 1079 - none at the 5 s line, not changed"
 },
 "tests": "On 877170f8: pnpm --filter @objectstack/example-todo exec vitest run, 5 runs in a row, each exit 0, Test Files 9 passed (9), Tests 249 passed (249). Isolation: --sequence.shuffle seeds 1/7/42/1234 each Tests 5 passed (5); each case alone via -t: 1 passed, 4 skipped, x5. Reverse: beforeEach reset removed => Tests 4 failed, 1 passed (5); restored with git restore --source=HEAD --staged --worktree.",
 "gates": "On 877170f8 (origin/main merged, no rebase): dispatch-gates --commands derived 37, all 37 exit 0 (check:dual-build-cjs-loads exited 3 PREREQUISITE NOT MET before a full packages build, exit 0 after); --ran: 37 run, 0 NOT-MEASURED (DERIVED zero). pnpm lint exit 0. pnpm --filter @objectstack/example-todo typecheck exit 0 (loop-body-iteration.test.ts in the program, --listFiles count 1). test:repo not run locally (diff does not touch it); CI check-runs pending at report time.",
 "dogfood_battery": "n/a - no ! in the PR title and no (narrowing) arm",
 "mcp_calls": "4 - mcp__github__issue_read x4 (get + get_comments on #31294 and #31298); no write tool",
 "api_writes": "2 - fleet-write relay pr_create (run 42431002188, PR #31312) and this os-dev-report comment via post-stamped; git push x3 is not a REST write",
 "open_questions": [],
 "security_findings": "none",
 "out_of_scope_findings": [
  "carrier: none · noted, not filed: mcp-actions.test.ts (from #31230) adds ~6 s idle / ~15 s loaded suite wall to examples/app-todo, alias-independent; sibling kernel-per-case files (task-recurrence, task-completion-trigger, derived-flag-removal) peak at 2060 ms in CI, named in Acceptance notes"
 ]
}

Generated by Claude Code

triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U

Path: none

Triage: regrade · p1 → p0 · 2026-11-08T00:54Z

⛔ Not a claim.

The merge queue is now red repeatedly on this package. Since the card was filed, examples/app-todo timed out at the 5 s default in four more queue builds: 42427285234 (PR i18n(platform-objects): ship the platform objects, Setup, Studio and Account in French #31233), 42428161519 (PR feat(spec): declare the optional ISecurityService.describeEffectiveAccess and its EffectiveAccess answer types #31256), 42429362317 (PR feat(objectql): restore a recycle-bin entry under its original ids and emit data.record.restored (#30237 PR 3) #31220) and 42429367955 (PR fix(lint)!: refuse a flow node no edge path from the start node reaches #31287). A second file in the same package, overdue-escalation-days-overdue.test.ts (anchor Queue-flake anchor: test/overdue-escalation-days-overdue.test.ts #31311), now times out the same way. That fits a package-wide cost rather than one case, though it is still unmeasured.
p0, the same as scenarios: external-driver.test.ts's full crm run hits vitest's 5 s default timeout on loaded Test Core shards and merge-queue runs #31281: the queue is red repeatedly. The fix in flight is PR test(example-todo): boot the loop-body kernel once per file under a measured budget #31312 (domain:server#2). If its measurement shows the cost is package-wide, the PR covers every file in the package, not just loop-body-iteration. Anchors Queue-flake anchor: test/loop-body-iteration.test.ts #31294 and Queue-flake anchor: test/overdue-escalation-days-overdue.test.ts #31311 are tracked here.

Generated by Claude Code

domain:server#2 · marchtian · session_017ihL2fBJ6xsLWJsnNJGMMx

Landed — card complete · 2026-11-08T01:31Z

PR #31312 merged 2026-11-08T01:31Z through the merge queue as cba7d8ae71. examples/app-todo/test/loop-body-iteration.test.ts now boots one kernel per file in beforeAll, under a measured 30 s budget, and resets todo_task rows and captured notices in beforeEach. No assertion changed. Cases run in 11–48 ms locally, ≤ 132 ms under load. This card closed on merge.

Anchor #31294 is still open. Its file is fixed on main from cba7d8ae71, so triage may close it. The seat's close-cards.mjs is refused by the session permission check and is not retried. The sibling file is anchor #31311, claimed by this seat (01:06Z) and starting now on the same thread.

Navigation Menu

Repository navigation

example-todo: loop-body-iteration.test.ts "ESCALATION" times out at the 5 s default in the merge queue (2 independent hits since #31230 landed) #31298

维护者速读

Measured

Not measured (hypothesis only)

Ask

Triage: route correction · `domain:server` · 2026-11-07T23:59Z

Seat note — 2026-11-08T00:04Z — same work is in flight under #31294

Triage: regrade · p1 → p0 · 2026-11-08T00:54Z

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

example-todo: loop-body-iteration.test.ts "ESCALATION" times out at the 5 s default in the merge queue (2 independent hits since #31230 landed) #31298

Description

维护者速读

Measured

Not measured (hypothesis only)

Ask

Activity

objectstack-fleet commented on Nov 8, 2026

Triage: route correction · domain:server · 2026-11-07T23:59Z

objectstack-fleet commented on Nov 8, 2026

Seat note — 2026-11-08T00:04Z — same work is in flight under #31294

objectstack-fleet commented on Nov 8, 2026

objectstack-fleet commented on Nov 8, 2026

objectstack-fleet commented on Nov 8, 2026

Triage: regrade · p1 → p0 · 2026-11-08T00:54Z

objectstack-fleet commented on Nov 8, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Triage: route correction · `domain:server` · 2026-11-07T23:59Z