Test site: risk-appetite-test.triodian.com

A separate copy of the app for trying changes before they reach the live engagement. It has its own Worker (risk-appetite-test) and its own database (risk-appetite-test-db), so nothing done there touches the live engagement. A striped banner across the top of every page says "Test site · not the live engagement".

One-time set-up (about 15 minutes)

1. Create the test database. Cloudflare dashboard → Storage & databases → D1 → Create database → name it risk-appetite-test-db. Copy its database ID and replace REPLACE_WITH_TEST_DATABASE_ID in wrangler.jsonc (or send the ID to whoever maintains the app). The tables are created automatically on first use.

2. Protect the address with the same sign-in. Zero Trust → Access controls → Applications → the Risk Appetite application → Configure → add a public hostname risk-appetite-test.triodian.com. Adding it to the same application keeps the same sign-in, the same MFA and the same AUD tag, so the app needs no other settings. Do this before the first deploy, so the address is never open.

3. Connect the test Worker to GitHub. Workers & Pages → Create → Import a repository → Triodian/risk_appetite.

   • Worker name: risk-appetite-test
   • Production branch: test
   • Deploy command: npx wrangler deploy --env test The risk-appetite-test.triodian.com sub-domain and its DNS record are created by Cloudflare on the first deploy, because the route is a custom domain. Nothing needs adding in DNS by hand.

4. Create the test branch in GitHub from main.

Day to day

• Changes go to the test branch first; Cloudflare deploys them to the test site.
• When they have been checked there, merge test into main; Cloudflare deploys the live site.
• The first person to sign in to the test site who is listed in BOOTSTRAP_ADMINS (lib/config.js) becomes its administrator. Add test users in its Admin, separately from the live site.
• To start the test site again from empty, delete and re-create the test database.