`meetings/2026-05-01_173651-morning.md`

54 lines
# Daily Stand-up — Fri 2026-05-01 17:36 PDT

---

### 1. Carryover from last meeting

- ⏳ `pd-api` — still absent from PM2. Not killed-intentionally, not restarted. Fate unknown.
- ⏳ `am-recrawl` — stopped, 90 restarts. No diagnosis yet. Hard blocker for anything downstream.
- ⏳ `sf-orchestrator` — still at 7 restarts in fleet data. No investigation result.
- ⏳ `george-gmail` — 3 restarts at 5h uptime. OAuth suspected, unconfirmed.
- ⏳ Directory decision — doctors + lawyers on one platform or two. Pre-scale window closing.

---

### 2. Overnight (last 12 hours)

**Crashes / instability:**
- `am-recrawl` — stopped, 90 restarts. No new data; still the priority wound.
- `animals-viewer` — 30 restarts, technically online. The 14:31 community fix has not stabilized it; restart count is high.
- `visual-factory-orchestrator` — 16 restarts. The prefetch/socket fix landed 18h ago but this restart count contradicts "stable." Fix may not be holding.
- `gemini-catalog-tagger` — 7 restarts in only 2h uptime. Fast crash loop; relatively new entrant to the problem list.
- `sf-orchestrator` — 7 restarts, still undiagnosed.

**Long-running jobs:** `am-recrawl` stopped — nothing crawling. All enrichment agents (full-monty, gap, update-sku, silas-sku) show 0 restarts / 32h uptime — stable and running.

**External signals:** No remote commits from collaborators. git activity is local only.

---

### 3. Today's plan + agent assignments

- `Steve / debugger-agent` — pull am-recrawl logs, identify crash-loop root cause — nothing downstream (recrawl feeds enrichment) — commit or restart with a fix within 24h
- `Steve` — decide pd-api fate: `pm2 list | grep pd-api`, then restart or formally retire — it's been a ghost since at least 14:47 — clear yes/no
- `pm2-crash-watcher` — deep triage on gemini-catalog-tagger (7 restarts / 2h) — fast crasher, new entrant — root cause identified
- `debugger-agent` — check visual-factory-orchestrator logs; 16 restarts says the prefetch fix didn't hold — success = either confirmed-stable or a follow-up patch committed
- `Steve` — confirm george-gmail OAuth credentials are valid — 3 restarts in 5h is the signature — credentials verified or rotated
- `Steve` — write the directory decision (one platform or two for doctors + lawyers) — deadline for this is now — decision in writing

---

### 4. Blockers and decisions Steve owes

- **am-recrawl root cause** — needs log pull before anyone can fix it
- **pd-api fate** — restart or retire; has been silent since at least early afternoon
- **george-gmail credentials** — OAuth check is a 2-minute task but it's blocked on Steve running it
- **Directory decision** — lawyers + doctors platform scope; every day without this is scope drift risk
- **visual-factory stability** — if the prefetch fix isn't holding, a second patch decision is needed

---

### 5. One thing to think about

`animals-viewer` is online but at 30 restarts — the community fix committed 3h ago may have reduced the *rate* but the process is clearly still unstable. Is the fix incomplete, or is there a separate crash path (memory leak, external dep) the debate-driven patches didn't address? Worth one log read before calling it resolved.