Hi Railway team,

Approximately 1 week ago a regression was introduced where the first

request to a sleeping serverless container always fails with a CORS error.

This was NOT happening before, the same code and config was working fine.

What's happening:

When a serverless container is sleeping, a browser sends an HTTP OPTIONS

preflight before the real request. Railway's queuing proxy answers that

OPTIONS request immediately (before the container is up) with a response

that has no Access-Control-Allow-Origin header. The browser then blocks

all subsequent requests with:

"Response to preflight request doesn't pass access control check:

No 'Access-Control-Allow-Origin' header is present on the

requested resource."

On a page refresh (container now warm), it works perfectly every time.

The second request in the same page load also always works.

My setup:

- FastAPI (Python) backends

- React/Vite frontend hosted on Vercel (https://app.studybuddyai.ca)

- Serverless enabled on Railway

- CORS middleware is correctly configured with allow_origins set to the

frontend domain

- Healthcheck path /health is configured in railway.json

- healthcheckTimeout: 60

What I've already confirmed:

- CORS middleware is set up correctly (works on warm containers)

- Healthcheck is configured, this doesn't help because the healthcheck

only runs on deploy, not on wake-from-sleep

- Issue happens across all serverless applications

- This was NOT happening ~1 week ago with the same code and configuration

Request:

Please either:

1. Have Railway's cold-start proxy pass through CORS request headers and

mirror them in the response (e.g. reflect Origin as

Access-Control-Allow-Origin on queued responses), OR

2. Advise what changed on Railway's infrastructure ~1 week ago that caused

this regression

This is causing a broken experience for all users who haven't visited

the page recently. Urgently need this resolved.