Project: ai agent dashboard
Service: @aiagent/api
Public domain: aiagentapi-production.up.railway.app
Region: asia-southeast1 (Singapore)
Plan: Hobby (Free)
Location: Taipei, Taiwan

SYMPTOM
All HTTP requests to my service consistently take 12-16 seconds. 
Problem started after deployment on 2026-04-25. Before this, latency 
was normal (~300ms).

Latency test (5 consecutive requests to /api/products):
- Request 1: 12.35s
- Request 2: 12.20s
- Request 3: 15.21s
- Request 4: 12.21s
- Request 5: 12.23s

The 12-second floor is extremely stable, which suggests a timeout 
or connection retry in the edge layer rather than variable network 
congestion.

EVIDENCE THIS IS NOT AN APPLICATION ISSUE
1. psql direct connection to my Postgres (same project, same region) 
   completes in 0.7s — DB is healthy
2. Non-existent 404 routes take 12s — requests never reach my app
3. /api/auth/login (which does NOT hit DB) also takes 12s
4. Application logs show NestJS started cleanly, all routes mapped, 
   no errors or warnings
5. CPU/memory well within Hobby plan limits
6. Nothing changed in global middleware or CORS config in this deploy 
   (git log confirms main.ts unchanged since scaffolding)

RESPONSE HEADERS (sample request)
- X-Railway-Request-Id: mY-z_ISKTiWOzthLacI7Nw
- x-railway-cdn-edge: fastly/cache-nrt-rjaa8190038-NRT  (Tokyo Fastly)
- x-railway-edge: railway/asia-southeast1-eqsg3a         (Singapore Metal)

QUESTION
Why is my request path going through Fastly Tokyo (NRT) before 
reaching Singapore Metal edge? Is there a misrouting at the CDN layer?

This pattern appears similar to previously reported APAC edge 
routing issues:
- https://station.railway.com/questions/metal-high-latency-singapore-b7548849
- https://station.railway.com/questions/server-latency-and-slow-response-times-991479b8

Could the Railway team investigate edge routing for my domain?

Thanks.