Daily Writeup — 2026-05-29 (Friday · MGP mid-day FAIL day 2 · Monthly-reports rebuild + new TOD lane · AMS prod testing burst)

Friday · Generated by nightly-project-manager · ← All Todos · ← May 28
0
Shipped
0
Moved Forward
0
Blocked
Mid-day MGP fails for the second consecutive day, but morning + recap both delivered cleanly — meaning the cron family of failures is narrowed to the mid-day editor specifically (not the whole notify path). Auto-pipeline at 16:30–16:46 ET ran the biggest monthly-reports refresh of the week: full param-surface regen for MSFT/SNDK/ORCL + ALL_SYMBOLS_COMBINED, plus a brand-new Time-of-Day (TOD) lane with six time buckets (open 9:30–10:00, early 10:00–10:30, mid-AM 10:30–11:30, midday 11:30–13:00, mid-PM 13:00–14:30, late 14:30–15:55) + tod_comparison.png + tod_sd_expansion.png. AMS prod got real testing activity 19:21–19:24 ET across _auth, app, cockpit, _api/cockpit.

TODAY — Top 3 (for Sat 2026-05-30)

  1. Diagnose the MGP mid-day cron specifically. Now two consecutive days where morning + recap deliver but mid-day silently fails (5/28 broke recap too; 5/29 only broke mid-day). The bug is narrower than "the whole MGP pipeline" — it's in the mid-day-edition path: scheduled task, gen_mgp_midday.py, the lunch-time renderer, or the SMTP call from that specific runner. Inspect the WTS scheduled-task definition and 5/27 → 5/28 → 5/29 mid-day call traces.
  2. Review the new TOD param-surface lane. Six intraday time buckets each with heatmap + surface PNGs at /db_reports/james-monthly-reports/vwap-bands/param-surface/tod_{heatmap,surface}_{open_930_1000,early_1000_1030,mid_am_1030_1130,midday_1130_1300,mid_pm_1300_1430,late_1430_1555}.png + comparison + SD-expansion plots. First time the param surface is sliced by time-of-day — needs eyes on whether the verdicts agree across buckets or split cleanly.
  3. Decide on the MGP v2 phase1_preview (deferred from 5/27 #2). The audit response is shipped at /mgp/v2/phase1_preview.html; today's mid-day failure is a different problem in a different file, so the audit-fix work is unblocked from delivery diagnostics. Promote or keep parallel.

What shipped

What moved forward

- /db_reports/james-monthly-reports/index.html + may-2026/index.html + vwap-bands/index.html + vwap-bands/param-surface-db/index.html + vwap-research/index.html all regenerated - vwap_param_{heatmap,surface}_{all_symbols_combined,msft,sndk,orcl}.png regenerated - NEW TOD lane — 6 time buckets × 2 plot types (heatmap + surface) + tod_comparison + tod_sd_expansion = 14 new PNGs first appearing today

What's blocked

Files touched