Hi Railway support,

I'm a Pro-tier user. My web service has been intermittently unreachable

for ~4+ hours today (June 22, roughly 14:00–18:00 UTC and ongoing).

Project ID: XXX

Service: the web service (Python/Flask via Nixpacks, public domain

proxied through Cloudflare on port 8080)

Region: US-West (a US-East replica was just added but wasn't active

when symptoms started)

Symptom:

• curl <my service's public URL>/health (a static "ok" endpoint)

  returns HTTP 000 — no TCP/TLS handshake completes within 10–15s.

  Same for /api/* paths.

• The Railway dashboard shows the service as Online the entire time.

  Internal Railway health checks aren't reflecting user-visible reality.

• Pattern is flapping: sustained down windows up to ~9 min (verified

  by 20 consecutive HTTP 000s spaced ~28s apart), brief recoveries to

  HTTP 200 (slow, 3–4s TTFB), then back to HTTP 000 within seconds.

• Static assets recover slightly faster than API paths but both

  ultimately hit the same gunicorn workers.

Ruled out on my side:

• App code (works fine during recovery windows)

• Cloudflare config (unchanged for weeks; would produce a Cloudflare

  5xx page, not HTTP 000)

• Client network (curl from independent residential ISP sees the same

  HTTP 000)

• Latest deploy shows "Deployment successful"; no push triggered the

  symptoms — they predate it

Your acknowledged incident (dashboard banner) mentions only "logs and

metrics may be slow to load." Request serving on at least my project

also appears affected — this looks broader than what's been posted.

Could someone check whether US-West nodes serving this project are

healthy and whether the gunicorn workers are actually receiving

traffic?

Thanks