← back to Claude Codex Meeting
meetings/2026-05-12_090001-morning.md
56 lines
# DW Morning Stand-Up — Tue 2026-05-12 09:00 PDT
---
## 1. Carryover
⏳ **dw_admin PG rotation** — Lane A runbook ready, never executed. Exposure now spans 6+ stand-ups.
⏳ **Gemini keys #1 and #2** — still hardcoded. No secrets activity overnight.
⏳ **Freddy port rotation** — soak window (post-Patty 24-48h) has long passed. Runbook ready.
⏳ **Doctor `apply-templates.js` decision** — Day 13. Lawyer pipeline still gated on Steve's yes/no.
⏳ **glassbeadedwallpaper diagnosis** — was 20 restarts at EOD; now 35. No log investigation run.
⏳ **Ventura-corridor sunset** — absent from pm2, no decision.
✅ **Designer-Wallcoverings archive-guard** — landed overnight; archive-guard wired into 5 cron-driven Shopify activators (15h ago).
---
## 2. Overnight
**Crashes / instability:**
- ⚠️ **wallco-ai** — 72 restarts. Was 11 at evening wrap; climbed dramatically overnight. Top priority log pull this morning.
- ⚠️ **whatsmystyle** — 32 restarts. Not flagged in yesterday's plan; needs a first look.
- ⚠️ **glassbeadedwallpaper** — 35 restarts (up from 20). Sixteenth consecutive morning without a fix.
- ⚠️ **pitchvideo** — 15 restarts. New entrant; no prior diagnosis.
- ⚠️ **thesetdecorator** — 14 restarts. Needs log check.
- **video-gallery, contact-mailer-ideas-viewer, cncp-failures-mockups, cncp-failure-collector** — 4–8 restarts each; chronic, low-urgency cluster.
**Long-running jobs:** No scraper or ingest pipelines show new overnight commits. decade-wallpaper fleet complete (32h ago). animals-ingest running, 1 restart / 34h — routine.
**External signals:** No collaborator commits. All recent git activity is Steve's own work. No Codex quota data visible.
---
## 3. Today's Plan
- `debugger-agent` — pull pm2 logs for **wallco-ai** and identify crash root cause — 72 restarts is an active loop, not passive — success: root cause named + patch or stop decision
- `debugger-agent` — pull pm2 logs for **glassbeadedwallpaper** — Day 16 undiagnosed — success: specific error identified, fix committed or process stopped
- `debugger-agent` — first look at **whatsmystyle** (32 restarts) and **pitchvideo** (15) — catch new crashers before they compound — success: triage verdict (fix / stop / deprioritize) for each
- `Steve` — **Doctor `apply-templates.js` decision** — yes or no — lawyer pipeline has been gated 13 days — success: logged decision, pipeline either unblocked or formally paused
- `secrets-manager` — execute **Gemini key #1 and #2 mint + route** via `/secrets` — open exposure — success: both keys in secrets registry, hardcoded values replaced
- `Steve` — **dw_admin PG rotation** — run Lane A runbook (`pg-rotation-runbook.md`, 32 min) — six stand-ups of open exposure — success: new password in place, all 14 hardcoded references patched
---
## 4. Blockers Steve Owes
- **Doctor `apply-templates.js`** — yes/no. Has been blocking lawyer pipeline for 13 days.
- **Gemini keys** — Steve must mint at aistudio.google.com and paste via `/secrets`. Agents can't do this step.
- **dw_admin rotation** — Steve must authorize and initiate the 32-min runbook. Agents can execute once authorized.
- **Ventura-corridor** — fix or formally sunset. Currently a silent gap in pm2 with no owner.
---
## 5. One Thing to Think About
**wallco-ai jumped from 11 → 72 restarts in 16 hours** — that's a crash loop, not normal churn. If wallco-ai is feeding any active pipeline (visual factory, SKU enrichment, Shopify writes), those jobs may be silently failing. Worth knowing what wallco-ai owns before the log pull, so you can assess blast radius.