Health Checks

Endpoints

Endpoint	Purpose	Expected response
GET /api/v1/health	Liveness check for backend service	200 with {"status":"OK"}
GET /	Base API availability check	200 with API metadata JSON

Monitoring signals

• Health endpoint availability (/api/v1/health success rate).
• API error rates (especially 5xx, 429, 401 spikes).
• Latency trends for core endpoints (/urls/public, /auth/login, /urls/my-links).
• Backend process uptime and restart frequency (Railway service events).
• Database connectivity and query failures (Prisma/PostgreSQL errors in logs).

Alert thresholds

Metric	Warning	Critical
Health endpoint failure rate	> 1% over 5 min	> 5% over 5 min
5xx error rate	> 1% over 5 min	> 3% over 5 min
P95 latency (/urls/public)	> 700ms	> 1500ms
Railway service restarts	>= 2 in 15 min	>= 5 in 15 min
DB connection failures	>= 1 recurring event	sustained failures > 2 min

These thresholds are starting points for portfolio-scale traffic and can be tuned with real usage.

Runbook

1. Confirm backend is reachable:
   • GET https://api.korta.click/api/v1/health
2. Check recent deploy and service status in Railway:
   • Running / Healthy
   • no route/path misconfiguration (/ public route, /api/v1/health health path)
3. Inspect backend logs:
   • Prisma connection errors
   • CORS/OAuth failures
   • surge in 429 or 5xx
4. Validate environment variables:
   • DB, JWT, CORS, OAuth, Resend
5. If incident persists:
   • rollback to last known good deploy
   • rotate compromised secrets if relevant
   • redeploy and verify /api/v1/health + critical flows (shorten, login, OAuth).