fix(telemetry): classify remaining expected failures in Sentry

未关闭
#50 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
4/5
预计耗时
3-5 天
新手友好度
32/100
Issue 类型
缺陷
描述清晰度
基本清楚
活跃度
活跃
技术栈
typescript
领域
observability

调研方向

Start with docs/telemetry.md and tests/telemetry.test.ts to see how expectedOutcome is defined and asserted, then audit each named Sentry group against the code that emits it, e.g. src/server/baguette-runtime.ts:77, src/ui/simulator-panel.ts:868, and the serve-emu/UI discovery paths. Check the 5 s simctl launchctl list deadline before filtering device-apps timeouts. Done means every listed failure has a classification, metric, or sample decision and the telemetry docs and tests match.

由索引模型根据 Issue 内容生成。

描述

enhancement

Commit 44f1b66 added expectedOutcome, but these user-environment and host-lifecycle failures are still not classified, so they keep being reported as errors. Each one should map to an expected outcome, be counted as a metric, or be sampled.

Toolchain missing (unsupported_platform or a new toolchain_missing)

  • Baguette Swift runtime libraries (src/server/baguette-runtime.ts:77): 3T, 2V
  • spawn /usr/bin/xcrun ENOENT: 9S, likely also 8Q (ios.devices.discovery, 69 users)
  • devicectl "requires Xcode 27": 4E, 21

Device-apps discovery deadlines

  • termination === "deadline" is deliberately left unclassified: 2N (ios_simulator_apps, 108 users), 3W, 61, 4V, 3C
  • Before filtering, check whether the 5 s simctl launchctl list limit is too tight. Consider raising it to about 15 s.
  • The same applies to UI discovery timeouts: UI-P

Physical iOS connection state (device_unavailable)

  • ios_physical_* causes connection_failed, connection_timeout, service_open_failed, service_open_timeout, query_timeout: 4B, 2W, 43
  • "iPhone no longer connected": 7P
  • "Connect the paired iPhone": 22
  • "adapter closed": 7V

Other server errors

  • Processes killed by signal (process_signalled → cancelled): AD, AC
  • adb timeouts and offline devices in serve-emu (device_unavailable): 7
  • ServeEmu.device: 1Q
  • Screenshot HTTP 400 (device_unavailable): 2T
  • Display FPS requires Android 12 (unsupported_platform): 9Y; FPS monitor already exists (invalid_input): 9F
  • "no accessibility data": 39, R. Also route plugin.ts:321 through describeUi so it gets the existing restart recovery.
  • tvOS or no-bezel simulators, definition.json 404 (unsupported_platform, fall back to no bezel): 29, 8X
  • Generic TimeoutError on tool deadlines (count as a metric, keep sampled): 2X, 36, X
  • app_flow "No package.json found" (invalid_input): A2

UI / host lifecycle

  • host_unavailable / session_closed outcomes:
    • UI-15, "app-server process is not available"
    • UI-12, "Transport closed"
    • UI-2X, "MCP client is shut down"
    • UI-2T, "composer unavailable"
    • UI-1F, "MCP app identity changed"
  • Best-effort cleanup calls that time out (*_close, *_reset, mobile_check_plugin_update, mobile_logs_close) should be counted as metrics, not captured as exceptions:
  • Small fixes while you're in there:
    • Guard setPointerCapture (src/ui/simulator-panel.ts:868, UI-2W)
    • Handle EPIPE on writes to a closed helper in resources.read (SERVER-A4)

Update docs/telemetry.md and tests/telemetry.test.ts to match.

主要语言
C++
星标
243
派生
21
平均合并
15 小时 6 分钟
30 天内合并 PR
57

环境准备

  • 没有 Dockerfile 或 Docker Compose 文件
  • 没有 Pull Request 模板
  • 阅读贡献指南

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

callstackincubator/codex-mobile-dev-plugin 的其他 Issue

查看 callstackincubator/codex-mobile-dev-plugin 的全部 Issue

相似的 Issue

更多 C++ Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。