`meetings/2026-05-06_090005-morning.md`

57 lines
# DW Morning Stand-Up — Wed 2026-05-06 09:00 PDT

---

## 1. Carryover

1. ⏳ **george-gmail OAuth rotation** — restarts climbed 9 → 12, uptime now 1h. No rotation landed overnight. Actively failing.
2. ⏳ **ventura-corridor crash loop** — restarts 20 → 44 overnight. Got significantly worse, not better.
3. ⏳ **Doctor apply-templates.js go/no-go** — 4 days since merge, no Steve sign-off. Lawyer Iter 46 also stalled behind same gate.
4. ⏳ **commerce-claw root cause** — restarts 11 → 39. Still uninvestigated.

---

## 2. Overnight

**Crashes / instability — fleet is NOT stable:**

The most important new signal: **8 wallpaper-niche sites all sitting at exactly 18 restarts** (agedwallpaper, textilewallpaper, pastelwallpaper, mylarwallpaper, architecturalwallcoverings, glitterwalls, greenwallcoverings, silkwallcoverings). Identical counts = shared root cause — likely a common module, shared Kamatera port block, or single broken dependency. This is one problem, not eight.

- `ventura-corridor` — 44 restarts (up from 20). smb-builder deployed to same Kamatera host 35h ago — port collision is still the lead suspect.
- `commerce-claw` — 39 restarts (up from 11). Cause still unknown.
- `national-paper-hangers` — 34 restarts (up from 4). Steep overnight climb, no prior context.
- `smb-builder` — 23 restarts. Fresh Kamatera deploy 35h ago, now crashing — migrations 002–008 may have a runtime issue.
- `george-gmail` — 12 restarts, 1h uptime. OAuth token expiry.
- `dw-universe` — 11 restarts. No prior flag; needs log check.
- `fleet-tv` — 8 restarts. No prior flag.

**Long-running jobs:** No overnight scrapes flagged. pd-crawler/pd-dedupe ticking normally (restarts=3/2). Pre-debate baselines committed across most projects 2 days ago — no new overnight ingest activity visible.

**External signals:** `business-claw` committed 43 min ago ("Launch comms drafts + og-cover.png raster + linter brand polish") — active work in progress right now.

---

## 3. Today's Plan

1. `Steve` — rotate george-gmail OAuth credentials — 12 restarts/1h uptime = actively broken, blocks email functions — success: restarts reset, uptime holds >2h
2. `debugger-agent` — triage the 8-site wallpaper cluster (agedwallpaper + 7 peers) for shared root cause — identical restart counts = one fix needed — identify common dependency or port conflict, single commit resolves all 8
3. `debugger-agent` — check smb-builder crash logs vs ventura-corridor for port collision on Kamatera — both deployed to same host in last 36h — one has to yield its port or be reconfigured
4. `commerce-claw` + `national-paper-hangers` — pull error logs for both — national-paper-hangers jumped 4 → 34 overnight which is alarming for an uncontextualized service — identify what it serves before deciding kill vs fix
5. `Steve` — greenlight or defer Doctor apply-templates.js — 4 days blocked, pd-crawler is ticking and building content with no template rotation — yes/no in 5 min
6. `Steve` — Gemini key rotation (two keys in open TODOs) — both were hardcoded across prod files — paste via `/secrets` after george-gmail is stable

---

## 4. Blockers Steve Owes

- **george-gmail OAuth** — no agent can rotate this; needs Steve's Google account access
- **Doctor apply-templates.js go/no-go** — blocking lawyer Iter 46 and doctor pipeline advancement
- **Gemini key #1 and #2** — minting blocked on Steve; rotation runbook is ready
- **PG dw_admin password** — Lane A runbook is at `~/Projects/_shared/docs/pg-rotation-runbook.md` (32 min), waiting for Steve's 32 minutes

---

## 5. One Thing to Think About

Eight wallpaper-niche sites crashed to identical restart counts overnight — if they share a process manager config or a single npm dependency that broke, one fix resolves all eight. But if they're actually separate Kamatera instances and this is a resource-starvation signal, the fleet may be hitting a ceiling. Worth knowing which before adding any more niche sites.