Documentation

Going live

Set your portal in one config file and deploy. The migrate step creates it, and the healthcheck fails the deploy instead of serving an empty site.

Taking your own portal to production takes one config file and a deploy. There is no seed to run and no admin to create first.

1. Set your portal

Everything that identifies your portal lives in packages/portal-config/src/portal.config.ts:

  • slug and name
  • domains, with your public domain first after any *.localhost names
  • compliance.brokerageName and compliance.brokeragePhone

The slug is set only here. The API compiles this file in, and the portal asks the API for it at GET /api/portals/current. See Configuration for the rest of the file.

2. Deploy

The migrate service runs npm run db:migrate, which applies migrations and then creates your portal (one Account and one active Portal) if it doesn't exist yet. Its log ends with one line:

✓ Portal smith-realty ready (created): Smith Realty · smithrealty.com · listings: hidden (MLS listings stay hidden until IDX approval)

On later deploys it says ready (existing, not changed). It only ever creates the portal. After that, edit the name, phone and so on from the owner dashboard, and a redeploy won't overwrite them. It never creates users, passwords or demo data.

Outside Compose, run npm run db:migrate as a release step before the API starts. The Deploying page covers Compose and Railway.

3. Check it

The portal's healthcheck is GET /api/health on port 3006. It is healthy only when the API serves your portal, so docker compose up --wait fails rather than serving an empty site.

Open <API_URL>/api/portals/slug/<slug>/readiness to see what's shown and why. When API_URL is set in the migrate service, the migrate log prints this URL for you.

When something is off

You see Cause Fix
Migrate log: portal.config.ts slug 'x' has no portal; this database has 'y' The slug in config changed after the portal was created, or the database belongs to another portal Put the existing slug back in portal.config.ts. A rename has to be deliberate: change the slug on the existing portal first. Nothing is created automatically.
Migrate log: Portal x exists but is inactive The portal was deactivated Reactivate it from the owner dashboard. The site returns 404 until then.
Portal /api/health 503: the API has no active portal… Migrate didn't run, or one of the two cases above Run docker compose run --rm migrate and read its last line
Portal /api/health 503: the API is unreachable at … The portal can't reach the API Check API_INTERNAL_URL and that the API is healthy

Demo data (db:seed:*) is for local evaluation only and refuses to run in production.

Next

MLS board setup — real listings, once your board approves the display.