#12-15s TTFB on all external requests — app is healthy (<1ms localhost)

2 messages · Page 1 of 1 (latest)

sand mica
#

Project ID: d922ba3e-9f07-4588-9fa1-3bb1749eca1e
Service: backend (ID: f3e2d484-d618-4e80-8b57-a5518fb48579)
Environment: production

Hi — looking for guidance on an edge/proxy issue.
Symptom: All external HTTPS requests to my service have a constant 12-15s TTFB (every endpoint, both HTTP/1.1 and HTTP/2).
Production users (mobile app with 15s timeout) cannot log in.

What I've verified:

  • Inside the container via railway ssh, localhost responses are <1ms — application is healthy.
  • App logs show requests processed in milliseconds.
  • Other services on my network (Vercel, Google, GitHub, Cloudflare R2) respond in <1s.
  • Two redeploys: first restored normal latency for ~5min, second had no effect.
  • Public IP resolves to 151.101.2.15 (Fastly).
  • Intermittent TLS handshake failures (curl exit 35) on the edge IP.
    I've already submitted a support ticket via help.railway.com with full details. Posting here in case anyone has seen this pattern recently or knows whether there's an ongoing edge incident affecting Fastly POPs near Korea.

Thanks!

wicked oracleBOT