| name | prod-up |
| description | Build and start the two production Docker containers (API + web app) for The Long Night on this host. Use when asked to bring up, start, boot, rebuild, or restart the production stack. |
Bring up the production stack
Builds and starts both prod containers with docker-compose-prod.yml:
HalloweenEventApi_prod (service halloween-api-prod, host port 5007 → HTTP) and
HalloweenEventWebApp_prod (service halloween-webapp-prod, host port 5009 → HTTPS, self-signed).
⚠️ This is the live production stack. On the Raspberry Pi (this deploy target) these are the
containers real players hit through the Cloudflare tunnel. The API uses the real Firebase node
(halloween-event) and real SMTP, and its lifecycle reconcile loop can send real emails near a
season boundary. To exercise the app without touching real data or mailing real people, use
/qa instead.
Normal deploys happen automatically: CI publishes a new image to GHCR and the in-repo
scripts/deploy-watcher.sh pulls it (see Deployment in CLAUDE.md). Run this skill for a deliberate manual
(re)start of the stack on this host.
Steps
-
Confirm the secrets are present (the compose file mounts/loads all three; without them the
build or boot fails):
ls -1 api.env app.env serviceAccountKey.json
If any is missing, stop and tell the user — on the Pi these live persistently in the repo dir
(gitignored); do not fabricate them.
-
Pull the published images and start (recreates both containers; -d = detached):
docker compose -f docker-compose-prod.yml pull
docker compose -f docker-compose-prod.yml up -d
This runs the current GHCR images (ghcr.io/cgoulart35/halloweenevent-{api,webapp}:latest) —
the same artifacts CI publishes and the watcher auto-deploys. For a full sync to origin/master
(code + images, like the auto-deploy), run sh scripts/deploy.sh. To rebuild locally from source
(dev/debugging only — the watcher replaces it on the next published image), use up -d --build.
-
Verify both are up:
docker compose -f docker-compose-prod.yml ps
Both services should show state running/Up. Then sanity-check the logs:
docker compose -f docker-compose-prod.yml logs --tail=20 halloween-api-prod
docker compose -f docker-compose-prod.yml logs --tail=20 halloween-webapp-prod
Healthy startup looks like each Flask/werkzeug app reporting it's serving (API on :5004
inside the container → published as 5007; web app on :5004 → 5009 HTTPS). Report any
traceback (common causes: a missing/blank required env var, or a bad FIREBASE_CONFIG_JSON).
Notes
- Quick restart without re-pulling (e.g. after nothing but a container hiccup):
docker compose -f docker-compose-prod.yml restart. Use the pull + up -d above to move to a
newer published image.
- Local URLs once up: API
http://localhost:5007, web app https://localhost:5009 (self-signed
cert — browser warning is expected).
restart: unless-stopped is set, so the containers come back on reboot/crash on their own.- To stop the stack, use
/prod-down. To inspect a running stack, use /prod-logs.
