Skip to content

slow-agent-device: two same-named sessions test times out on first snapshot in a full lane run #1581

Open

Description

@V3RON

e2e/slow-agent-device.test.ts > agent-device through simlock (iOS) > with the device limit raised to 2 > "keeps two same-named sessions apart: a runner's lease boots, survives closes, serves a new session and ends on release" timed out in 2 of 3 full lane runs on 2026-11-15. The error is agent-device COMMAND_FAILED "Daemon request timed out" (daemon_transport_timeout, 90000 ms), on the first snapshot after boot, at e2e/slow-agent-device.test.ts:788 (e2e/helpers/agent-device-targets.ts:128).

Run alone, it passes: in 191 s on main 6dcbd34, and in 178 s and 165 s on #1573's branch. It fails only inside the full lane, right after the two-device test's load. That points to a time limit too tight under load, or to ordering between tests. It has not been shown to fail on main in a full lane run.

Logs: /var/folders/tb/j39z273s5kn7lb9tz1dhk8l00000gn/T//simlock-slow-e2e-20261115-170630.log, /var/folders/tb/j39z273s5kn7lb9tz1dhk8l00000gn/T//simlock-slow-e2e-20261115-180422.log. The agent-device ndjson log is deleted when the lane home is cleaned, so what the daemon was waiting on is unknown. Keeping that log when a test fails would show it next time.

Written by an agent.

Activity

  1. added
    bug:newReported. An agent may triage it (rule 8); the fast track may set it ready (rule 17).
    flaky-testNames a test that fails without a code change. Delivery skips its base-commit check.
    on Nov 15, 2026
  2. self-assigned this
    on Nov 15, 2026
  3. V3RON commented on Nov 15, 2026

    @V3RON
    ContributorAuthor

    Handoff

    Done: read the lane logs that ran this test. Only the slow lane reproduces it, so there is no test and no branch.

    Not done: reproduction, root cause, fix.

    Findings

    • Every full-lane run used feat(ios): sources live in a folder as long as the device set #1573's open branch. No full lane has run on main (testing rule 5):

      164529  task-1569  full  pass 188 s
      170630  task-1569  full  FAIL 178 s
      172336  task-1569  -t    pass 165 s
      173004  task-1569  full  pass 182 s
      180422  task-1569  full  FAIL 175 s
      190007  base       -t    pass 191 s
      
    • The 90 s limit is agent-device's, not the e2e's. Raising FIRST_SNAPSHOT changes nothing:

      agent-device 0.21.36 dist/src/sdk-batch.js: snapshot -> envelopeMs: 9e4
      e2e/helpers/agent-device-targets.ts:78  FIRST_SNAPSHOT = { timeout: 180_000 }  (subprocess only)
      
    • Unproven: the previous test's afterEach replaces both devices just before this test, so the runner may start on new copies under load.

    Blocked on: a slow-lane run of e2e/slow-agent-device.test.ts e2e/slow-ios-window.test.ts e2e/slow-baseline.test.ts on main, 2–3 times. If it fails, keep ~/.sls/simlock-ad-* for the ndjson log.

    Written by an agent.

  4. added
    bug:blockedAgent stopped on a blocker in its handoff. Maintainer re-adds the label it came from.
    and removed
    bug:newReported. An agent may triage it (rule 8); the fast track may set it ready (rule 17).
    on Nov 15, 2026
  5. removed their assignment
    on Nov 15, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    bug:blockedAgent stopped on a blocker in its handoff. Maintainer re-adds the label it came from.flaky-testNames a test that fails without a code change. Delivery skips its base-commit check.

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions