Three structural patterns from the 2×2 caller×callee version matrix: callee version controls first-audio (−183ms vs −29ms caller), 4.11.3 adds +130–190ms init conn on both sides, audio quality only improves when both sides are upgraded.

2×2 heatmaps across 4 metrics: callee version controls first-audio (6× asymmetry), 4.11.3 adds +130–190ms init conn on both roles, MOS only improves with mutual upgrade. Raw data from roomid_match_7d_di, 19,751 PK_PK call pairs.

SDK-side in-call quality triggers: 14 active prompts across action (T01–05), informational (T10–23), and peer-inferred (T17/T19) categories. Includes priority tiers P0–P5, merge and suppression rules.

2 bugs ONLINED (VPQ-360 iOS crash, VPQ-268 no-ring) ahead of the 4.12 release; 2 Lighthouse stories Released (Meet + P2P KPI cards); FPM Drop Funnel analysis picked up.

Quiet day: COR-361 Voice Bot SME Lending at 70% (user-callback in UAT); kai.wang gateway/FPM work carried from 06-22, COR-372 Rust prod go-live still blocked on AppCenter config.

VPQ Sprint 24 — 6 tickets changed Mon Jun 22: VPQ-360 iOS BOTRTCV4 crash reopened then BUGFIXED (Firebase rolled back to 12.7); ML-VQC blocked on backend mlvqc_enabled flag; new color-space bug VPQ-364; V4 grayscale → In Testing.

COR Sprint 22 — activity Mon Jun 22 from kai.wang (gateway Phase 3 at 90% after clearing an AppCenter port connection-limit; FPM Data Insight 60%, 6 services in prod) and Hui Liu (Voice Bot SME Lending server 65%).

Sunday (UAE weekend): only one China-TZ commit landed — Zhao shipped ML-VQC low-MOS (ty=9) quality-event reporting on VPQ-116 across iOS+Android, builds green.

Sunday (UAE weekend): no COR activity for WenlinYuan or Pengxiang.Xie. Standing in-flight queue carried into the new week — 22 Platform Terms web bugs await review; change-phone-number backend and India 5th line still in dev.

Fri+weekend: 3 crash/bugs cleared (VPQ-340/356/314), Lighthouse-for-Server released (VPQ-343), iOS crash fix awaiting verify (VPQ-360), new 4.12 reconnect bug (VPQ-367), ML-VQC A/B + ty=9 progress (VPQ-116).

Fri+weekend: no Jira activity from WenlinYuan or Pengxiang.Xie (last touch Thu 06-18); in-flight items COR-329 / COR-353 carried forward unchanged.

Core COE standup covering last Friday + the weekend (2026-06-19 → 06-21). Whole COR team: kai.wang was the only owner with movement (all on Sun 06-21) — Rust TCP gateway Phase 1 Released, Phase 2 UAT-tested but prod blocked on Konghui's AppCenter config, Java migration ramped to 85%, Lucy/Lighthouse-for-Server Closed, FPM Data Insight live. WenlinYuan & Pengxiang.Xie no movement.

Meet 4.11.0 shipped an iOS-only SDK change; on iOS bad-call rate dropped 59.4%→31.9% while call success fell 78.7%→73.1%. Android (no SDK change) is flat across the same window — clean control.

7-day VoIP quality: bad rate 9.09% (-0.24% vs 4w baseline), QA & SD above threshold, dxblink6/bomlink6/fjrlink1 degrading, rating stable 4.38.

36 COR tickets: COR-329 change-phone backend 70%→90% (auth/security), COR-353 India 5th line blocked on ops IP binding, 22 Platform Terms web-bugs in review, COR-313 released.

v4 bad-call judgment on kVoipSummaryStat: hybrid engine + deliberate-action exclusion + two-layer dedup, inheriting v2's A/V/C+R skeleton. All thresholds TBD.

基于 kVoipSummaryStat 的 v4 bad-call 判定：混合引擎 + 主动行为排除 + 两层去重，继承 v2 的 A/V/C+R 骨架(中文版)。全部阈值待标定。

VoIP Sprint 24 — 12 of 27 tickets moved 2026-06-18: VPQ-334 travel/region reallocate Released (im-chatserver MR#169), push-token tracking VPQ-35/354 Closed, Abhishek's Lighthouse build push; ML-VQC green-tint (VPQ-116) still reproduces after fix.

Core COE — COR-329 change-phone-number backend jumped 70%→90% (kick-flow + auth hardening); COR-353 India 5th line still blocked on ops IP binding; COR-368/370 refined (To Do).

Standup covering 2026-06-17: 4 changed. VPQ-340 P1 crash regressed (QA reopened, re-fixed); VPQ-357 freeze-metric fix → Pending Confirm; VPQ-351 Voice Bot started; VPQ-116 H265 QP bug fixed.

COE covering 2026-06-17: COR-329 change-phone backend 70% (PayBy callback sig verify); COR-353 India 5th line started (15%, handed to ops); 4 new items planned (COR-368/369/370/371).

Standup covering 2026-06-16: 4 tickets progressed (commits/comments only, no status changes) — VPQ-357 V4 freeze-metric fix prototyped, VPQ-343 blocked on @kai.wang token, VPQ-270 dashboard needs Xinyong Ma tables, VPQ-116 Strategy B + self-test.

Core COE Sprint 21 kickoff (planned 2026-06-16): 31 items for WenlinYuan (27) + Pengxiang.Xie (4). Highlight: 22 Platform Terms web bugs in review; COR-313 Lucy configurator released. No commits/comments (planning day).

Weekend recap (Fri 06-12 → Sun 06-14): all activity on Friday; 7 changed, 2 BugFixed (VPQ-340 P1 crash, VPQ-268), 1 new bug (VPQ-356). Sat/Sun no activity.

Per-channel success funnel, failure reasons, and drop-off analysis — table layer plus the instrumentation TODOs needed to build it.

分渠道成功漏斗、失败原因、流失原因 — 表层设计 + 需要的埋点整改(中文版)。

Standup covering 2026-06-11: only 3 tickets moved — VPQ-343 To Do→In Dev (new deploy blocker), VPQ-270 blocker now owned by Xinyong Ma, VPQ-116 commit. Rest unchanged (sprint-creation day).

Sprint 24 standup: 14 changed, 5 silent; 2 high-risk (VPQ-340 P1 crash, VPQ-270 blocked); 5 stray tickets need sprint triage.

Real per-3h Kibana data (logs-footprint-voipstat-*, name=kVoipTinyStat, pp=AE_PH, 2026-06-01 → 06-08, 61 buckets). Dedicated lines (dxblink7 / sglink7) run RTT ~17 ms below their public counterparts (285/290 ms vs 303/306 ms); packet loss and answer rate (~88% state-2 across all routes) are on par. Dedicated lines carry ~7.5% of AE_PH volume. Keep/drop decision pending cost analysis. 4×4 Kibana-style grid; click any panel to expand. Charts embed data inline — render offline.

Daily trend of inviteDelay (callee, is_caller=0) and inviteToRing (caller, is_caller=1) over 2026-05-05 to 06-03. inviteToRing avg rose 2.3 → 38.8 s with a single-day spike of 87.3 s on 05/27; inviteDelay p95 stepped up to ~3,050–3,450 s across 05/27–05/31 (peak 3,450.7 s on 05/27). p50 flat throughout.

AE→AE corridor drilldown of inviteDelay (callee) and inviteToRing (caller) over 2026-05-05 to 06-03. inviteToRing raw avg rose 1.72 → 23.09 s while the <36000 s filtered avg moved only 1.72 → 2.74 s (rise is the ≥36000 s tail). inviteDelay filtered avg and p95 both peaked on 05/27 (438.7 s / 1,892.6 s). p50 flat throughout.

New-SDK grey rollout paused (expected to resume 2026-06-10); Lighthouse data cleaning finished today with v3 vs v4 now comparable; call-not-ring breakdown still missing in Lighthouse (TODO).

Full weekly report (CN/EN) 2026-05-18 to 05-24 · OP S22 closed + S23 kickoff · VoIP S22 mid · 46 parent rows.

iOS +2.31pp · Android +0.81pp · onset May 22 · peak May 27 · Silence Never Ring is the primary driver.

Sprint 23 · OpenPlatform 35 + VoIP Quality 11 items In Dev, grouped by component (Platform MP / Server / Onboarding / AI Lighthouse / SDK / ...), with weekly progress placeholders.

Sunday-only spike: bad-call rate hit 10.14% (+0.97pp vs prior 4-Sun avg of 9.17%). Cross-platform, cross-version, all 12 monitored corridors moved. Signature: mid-call audio freeze + long connecting-fail (>2s).

Open Platform Q2 OKR: Objective + 6 KRs across Speed / Scale / Partner pillars. Horizontal swimlane layout.

VoIP Quality Q2 OKR: reduce bad call 5%, resolve 5 field issues, PH/IN dedicated line optimization, 2 POCs (Zoom + Microsoft). 4 KRs across Quality / Network / Innovation pillars.

Sprint 22 · OpenPlatform 20 + VoIP Quality 19 items In Dev, grouped by component (Platform MP / Server / Onboarding / AI Lighthouse / SDK / ...), with weekly progress placeholders.

Full weekly report HTML (CN/EN tabs) for 2026-05-11 to 05-17. VoIP S21 wrap-up + S22 kickoff; OP S22 in progress. 29 by-parent rows, English default.

Android silent-ring rate hit 18.96% on 2026-05-11 — the single highest day in the trailing 90 days (+1.60pp vs 5/8 baseline). All 11 monitored corridors moved the same day. Real Grafana data. Investigation tracked in PB.

30-day evaluation (Apr 6 → May 6) of Zoom Video SDK Lite 2.4.6 as a P2P media layer. Paused 2026-05-07: POC measured 4.4–6.0 s joinSession; TestDevLab independently measured 5.6 s ringing on Zoom’s consumer app vs Botim 1.06 s. Three compounding blockers — connect time, Android minSdk 28 vs Botim 21 (~29K DAU lockout), and missing onQOSStatisticsReceived in Lite AAR.