Everything I could decide myself is decided. What's left needs
you. Each one carries the recommendation of the session that raised it — if you
agree, pick it and press Done. A model's opinion, where there is one, appears on the card
as a receipt line; cards with no receipt were never put to a model. "I'm not sure" is a real answer and becomes work for me.
You don't need to open the session — pressing Done sends your answer back and it picks up.
1work-order re-enrich sweep (propflow-appfolio-sync-prod-wo-enrichment): 1,074 open AppFolio-linked WOs re-queue daily (20h freshness < daily cadence) but one 900s Lambda drains ~190 (L4 enrich mean 3.1s = 92% of the 640s drain; Lambda already at max; the queued skip-the-DDB-scan follow-up buys nothing). 865 never-enriched Western Slope rows lead the global stalest-first queue, so Camellia's 21 (refreshed daily 09-12/13/14) got 0 today and breach 48h tomorrow. The budget-staleness alarm fires every morning until the shape changes — which shape?
In plain terms. Every morning we re-read each open work order from AppFolio so the 'Vendor on File' card stays current. Since yesterday's credential fix, Western Slope's ~1,050 work orders joined that morning queue — about 5 times more than one morning can get through (about 190) — and because they had never been read they went first, so Camellia's 21 got nothing today and will still be waiting tomorrow. This alert fires every morning until we change the shape. Do we split the run per customer and refresh Western Slope weekly, run it every 4 hours, or slow everyone to weekly?
The session raising this recommends: Split per AppFolio account; Camellia daily, Western Slope weekly Restores Camellia's daily refresh tomorrow with zero added AppFolio load (the 429 storm that parked the WO mirror on 09-12 is three days old; AppFolio's ceiling is ~28 req/min), matches the existing one-worker-per-account shape, and gives Western Slope's ~1,050 rows a cadence the budget can meet (~150/day) instead of an alarm every morning. Tried first, unsuccessfully: Read the run log (runId d61c1f2b): 1,074 queued / 188 enriched / 0 failed / 886 unreached = 865 never-enriched (all on the Western Slope org) + Camellia's 21 at 24.1h. Measured L4 enrich mean 3,144ms (n=188, 591s of the ~640s drain), so the DDB filter-scan follow-up is ruled out as a capacity fix. Checked the 3 prior mornings: Camellia 21 enriched / 12 vendors each day; the whole run took 65-177s because Western Slope failed fast on the credential bug fixed 09-14. Confirmed no open block carries this cause (b89411968 and b89304739 are resolved, different causes). Caveat: the WO status mirror has been parked since 09-12, so some of the 1,074 'open' rows may already be closed in AppFolio — true steady-state demand may be lower than 5.7x over budget. Did NOT re-run or hand-invoke the sweep (prod mutation) and did not change the queue ordering (designed on purpose).