Skip to content

supabase-store-runbook.md

Latest commit

 

History

History
69 lines (52 loc) · 2.28 KB

supabase-store-runbook.md

File metadata and controls

69 lines (52 loc) · 2.28 KB
Raw

SLL-R Supabase Store Runbook

SLL-R can persist orders and runtime demo merchants in Supabase Postgres through the PostgREST HTTP API (no SDK dependency). This is one of the durable store backends; see also the Vercel KV / Upstash REST option in env.example.

1. Create the tables

Run this in the Supabase SQL editor (Dashboard → SQL):

create table if not exists sllr_kv (
  key text primary key,
  value jsonb not null
);

create table if not exists sllr_index (
  index_key text not null,
  member text not null,
  primary key (index_key, member)
);

SLL-R talks to these tables with the service-role key, which bypasses row level security, so no RLS policies are required. Keep RLS enabled (the default) so the anon/public key cannot read or write order data.

2. Configure SLL-R

Set these environment variables (Supabase → Project Settings → API):

SUPABASE_URL=https://<project-ref>.supabase.co
SUPABASE_SERVICE_ROLE_KEY=<service-role-secret-key>

SUPABASE_SERVICE_ROLE_KEY is a secret — keep it server-side only (it is never sent to the browser). SUPABASE_KEY / SUPABASE_SERVICE_KEY are also accepted as fallbacks.

When configured, GET /health reports { "store": "supabase" }. Backend selection order is Supabase → Redis/KV → memory, so Supabase wins if both are set.

3. How state maps to the tables

  • sllr_kv holds one row per order (sllr:order:<id>) and one atomic demo merchant registry (sllr:demo-merchants:v1); value is the JSON document.
  • sllr_index holds the membership sets: the global order index, the per-merchant order index (sllr:order-ids:<merchantId>), and legacy demo merchant ids used only during migration. Inserts use resolution=ignore-duplicates, so re-adding a member is a no-op.

4. Verify

curl -s https://<your-sllr-host>/health
# {"ok":true,"product":"SLL-R","store":"supabase"}

Create an order, then confirm it survives a fresh instance by reading it back in a separate request:

curl -s -X POST https://<your-sllr-host>/merchants/raposa-coffee/orders \
  -H "content-type: application/json" \
  -d '{"userIntent":"iced latte in 10 minutes","deadlineMinutes":10,"maxSpendUsd":"10.00","paymentMode":"counter"}'
# note the order id, then:
curl -s https://<your-sllr-host>/orders/<ord_id>