Production Deployment Plan - Historical Netlify Baseline¶

Status: Historical reference; Netlify is no longer the live frontend host Date: March 12, 2026 Platform: Netlify (historical production baseline; rollback-only)

Update (March 12, 2026): Netlify deploys were restored and the application served successfully from the Netlify site, https://wait-time.ca was live with a valid Let's Encrypt certificate, https://www.wait-time.ca redirected to the canonical host, and production smoke passed against the canonical domain.

Runtime usage controls implemented: - /api/health polling reduced to every 5 minutes and only while tab is visible. - Read-heavy API routes now send short CDN cache headers (typically 2-10 minutes). - Admin, geolocation, and export routes are Cache-Control: no-store.

Direct-VPS note (March 13, 2026): the frontend is now live on the shared VPS and the active runtime path is documented in docs/operations/direct-vps-frontend.md. Treat this file as historical Netlify baseline context only.

Current State¶

What's Already Built¶

• ✅ Backend scrapers: Python with GitHub Actions cron (hourly schedule active)
• ✅ Frontend: Next.js 14 with SSR, now deployed on the shared VPS
• ✅ Database: Neon PostgreSQL (already in production)
• ✅ CI/CD: GitHub Actions for scrapers, maintenance, and monitoring

What Needs Deployment¶

1. No blocking deployment tasks remain
2. Ongoing release discipline via explicit [release] commits
3. Launch-material follow-through (LinkedIn post, screenshots, walkthrough)

Deployment Architecture Options¶

Option A: Static/SSG Hosting (Recommended for MVP)¶

Platforms: Vercel, Netlify, Cloudflare Pages

Pros: - Zero-config deployment for Next.js - Automatic builds on git push - Free tier generous (Vercel: 100GB bandwidth) - Built-in CDN and edge functions - Environment variable management in UI

Cons: - Vendor lock-in to some extent - SSR cold starts on free tier - Limited control over infrastructure

Best for: Quick launch, portfolio project, low operational overhead

Option B: Container Hosting¶

Platforms: Railway, Render, Fly.io, DigitalOcean App Platform

Pros: - More control over environment - Can run both frontend and backend services - Docker-based (portable) - Often includes database hosting

Cons: - Requires Dockerfile configuration - More setup complexity - Potentially higher cost - Need to manage container updates

Best for: Scaling beyond MVP, multiple services, specific runtime needs

Option C: Self-Hosted (VPS)¶

Platforms: DigitalOcean Droplets, Linode, Hetzner

Pros: - Complete control - Lowest cost at scale - Can run anything - SSH access for debugging

Cons: - Manual setup (nginx, SSL, etc.) - Requires devops knowledge - You manage security updates - No automatic scaling

Best for: Cost optimization, learning infrastructure, high control needs

Option D: Serverless Functions Only¶

Platforms: Vercel Functions, Netlify Functions, AWS Lambda

Pros: - Pay per execution - Auto-scaling - No server management

Cons: - Cold starts - Execution time limits (10s on Vercel free) - Not ideal for long scraper runs - Would need to refactor scraper architecture

Best for: Micro-services architecture (not recommended for current setup)

Historical Approach: Render (Superseded)¶

Platform: Render (container-based hosting)

1. Frontend on Render
2. Next.js support via Web Service
3. Free tier available (with limitations)
4. Auto-deploy from GitHub

5. Environment variables: NEXT_PUBLIC_MAPBOX_TOKEN, DATABASE_URL

6. Scrapers on GitHub Actions (already configured)

7. Runs hourly on the live GitHub Actions path
8. Connects to Neon PostgreSQL

9. No additional hosting cost

10. Database on Neon (already in production)

11. Free tier: 512MB storage
12. Auto-pause when idle

Total Cost: $0/month on free tiers (Render free tier has spin-down on inactivity)

Decisions Made¶

1. Frontend Hosting Platform¶

• Render (account created)

2. Custom Domain¶

• Use default Render domain (*.onrender.com) for now
• Custom domain may be purchased later

3. Alerts for Scraper Failures¶

• Pushover notifications (mobile push notifications)
• Required credentials: PUSHOVER_USER_KEY, PUSHOVER_API_TOKEN

4. Hosting Tier¶

• Free tier (instances spin down after 15 min inactivity)
• First request after spin-down will be slow (~30s cold start)
• Upgrade to paid ($7/month) later if needed

5. Monitoring/Analytics¶

• Enable Render metrics dashboard (optional)
• Set up Sentry error tracking (optional)
• Monitor via GitHub Actions + Pushover alerts for now

User Actions Required¶

Once decisions are made above:

Step 1: Configure GitHub Secrets¶

Add to repository Settings → Secrets → Actions:

Getting Pushover credentials: 1. Create account at pushover.net 2. Install Pushover app on your phone 3. Your User Key is on the dashboard after login 4. Create an Application → get the API Token

Step 2: Deploy Frontend¶

Instructions will depend on platform chosen (see platform-specific guides below).

Step 3: Verify Deployment¶

• Frontend accessible at public URL
• Map displays with hospital markers
• /methods page loads
• API routes returning data
• Scraper cron runs successfully

Platform-Specific Deployment Instructions¶

Render (Selected)¶

1. Go to dashboard.render.com → Sign in
2. "New" → "Web Service" → Connect GitHub repo
3. Configure service:
4. Name: waittime-canada (or preferred name)
5. Root Directory: frontend
6. Environment: Node
7. Build Command: npm install && npm run build
8. Start Command: npm start
9. Add environment variables:
10. NEXT_PUBLIC_MAPBOX_TOKEN = (Mapbox token)
11. DATABASE_URL = (Neon connection string)
12. Deploy

Notes: - Free tier instances spin down after 15 minutes of inactivity (first request may be slow) - Paid tier ($7/month) keeps instance running continuously - Deploys are release-gated in this repo; production deploy requires commit message containing [release] or [deploy]

If Vercel Selected (Alternative):¶

1. Go to vercel.com → Sign in
2. "Add New" → "Project" → Import GitHub repo
3. Set Root Directory: frontend
4. Add environment variables:
5. NEXT_PUBLIC_MAPBOX_TOKEN = (Mapbox token)
6. DATABASE_URL = (Neon connection string)
7. Deploy

Configuration file needed: frontend/vercel.json

If Netlify Selected:¶

1. Go to netlify.com → Sign in
2. "Add new site" → Import from Git
3. Build settings:
4. Base directory: frontend
5. Build command: npm run build
6. Publish directory: .next
7. Add environment variables (same as Vercel)
8. Deploy

Configuration file needed: frontend/netlify.toml

If Railway Selected:¶

1. Go to railway.app → Sign in
2. "New Project" → Deploy from GitHub
3. Add service: Frontend
4. Root directory: /frontend
5. Build command: npm run build
6. Start command: npm start
7. Add environment variables (same as above)
8. Deploy

Configuration file needed: frontend/Dockerfile (optional but recommended)

If Cloudflare Pages Selected:¶

1. Go to dash.cloudflare.com → Pages
2. "Create a project" → Connect to Git
3. Build settings:
4. Framework preset: Next.js
5. Build command: npm run build
6. Build output directory: .next
7. Root directory: frontend
8. Add environment variables (same as above)
9. Deploy

Note: Cloudflare Pages requires Next.js Edge Runtime compatibility

Risk Assessment¶

Post-Deployment Monitoring¶

After deployment, monitor for 24-48 hours:

1. GitHub Actions
2. Check scraper-cron runs hourly
3. Verify heartbeat-monitor passes hourly

4. Watch for failure notifications

5. Frontend

6. Test all pages load
7. Verify data displays correctly

8. Check browser console for errors

9. Database

10. Monitor connection count in Neon dashboard
11. Verify measurements are being inserted
12. Check retention cleanup / storage reporting runs as intended

Rollback Plan¶

If deployment fails or has critical issues:

1. Frontend: Redeploy previous version via platform dashboard
2. Scrapers: Disable GitHub Actions workflow temporarily
3. Database: No changes needed (data persists)

Next Steps After Deployment¶

1. Update README.md with live demo URL
2. Update LinkedIn post with actual screenshots
3. Monitor for 48 hours for stability
4. Consider custom domain setup
5. Begin planning Milestone 4 (Quebec expansion)

All Decisions Complete¶

All deployment decisions have been made. Ready to proceed with deployment:

1. ✅ Platform: Render (free tier)
2. ✅ Domain: Default *.onrender.com
3. ✅ Alerts: Pushover notifications
4. ✅ Tier: Free (with spin-down)

All decisions made. Ready for deployment.