a month ago
I have a production Next.js application deployed on Railway. The deployment is SUCCESS/RUNNING, but the Railway-generated public domain returns a Railway fallback 404 and requests never reach the application.
PROJECT
- Project: el-montazah-elegance
- Project ID: a5b51415-1289-45a0-a9ef-e929c9bd90f8
- Environment: production
- Environment ID: fa245cb0-1063-4499-a359-4aba82230249
SERVICE
- Service: web
- Service ID: 1466f4bd-c5b1-45a5-b1eb-2d04c2fc3643
DEPLOYMENT
- Deployment: 9f121639-6329-4fd7-9970-f621a5061293
- Deployment status: SUCCESS/RUNNING
- Runtime instance: 98ed87a6-812d-4ce9-be3c-e213649bcef5
DOMAIN
- Railway-generated domain: web-production-8f58.up.railway.app
- Domain ID: 4e655a85-1304-40ac-b06e-9ed78ff57bc4
- Target port: 8080
- Domain sync status: ACTIVE
APPLICATION
- Next.js 15.5.25
- Application listens on 0.0.0.0:8080
- Production build completes successfully
- .next output is valid
- routes-manifest.json and required-server-files.json are present
- 80+ application routes are included in the production build
PROBLEM
Requests to the Railway-generated domain return:
HTTP 404
x-railway-fallback: true
Server: railway-hikari
The response is Railway's fallback response rather than an application response.
Most importantly, there are ZERO HTTP request logs reaching the application, including when querying the active deployment. This indicates that requests are not being forwarded from the Railway edge/gateway to the running container.
I initially verified the application was listening on port 8080. The Railway domain was originally targeting port 3000, so the target was changed to 8080. The service was then restarted and a fresh deployment was created. The current deployment is still SUCCESS/RUNNING and listening on 0.0.0.0:8080, but the public domain continues to return x-railway-fallback: true / 404.
TROUBLESHOOTING ALREADY PERFORMED
- Confirmed deployment is SUCCESS/RUNNING.
- Confirmed Next.js starts successfully on 0.0.0.0:8080.
- Confirmed production build is valid.
- Confirmed the Railway-generated domain exists and is attached to the web service.
- Confirmed domain target port is 8080.
- Confirmed domain sync status is ACTIVE.
- Changed the domain target from 3000 to 8080 because the application listens on 8080.
- Restarted the service.
- Created a fresh deployment.
- Confirmed the fresh deployment is also SUCCESS/RUNNING.
- Confirmed the application still produces no HTTP request logs when the public domain is accessed.
- No Prisma migrations, database resets, db push, or database modifications have been performed.
REQUEST
Please investigate the Railway edge/gateway/Hikari routing and replica readiness for this domain and deployment.
Specifically, please verify:
- whether the running replica is considered healthy and routable by Railway;
- whether the healthcheck is passing;
- whether the domain is correctly bound to the active deployment/replica;
- whether there is an edge/Hikari routing issue preventing traffic from reaching the container.
The application and PostgreSQL database should NOT be modified as part of this investigation.
2 Replies
a month ago
This thread has been opened as a bounty so the community can help solve it.
Status changed to Open Railway • 28 days ago
0x5b62656e5d
Try deleting the URL and recreate it.
a month ago
I followed your recommendation to delete/recreate the URL and verified the active deployment again.
The issue persists.
Current deployment:
- Deployment: 9f121639-6329-4fd7-9970-f621a5061293
- Status: ACTIVE
- Next.js 15.5.25
- Listening on 0.0.0.0:8080
- Ready successfully
Current Railway-generated domain:
web-production-8f58.up.railway.app
The domain targets port 8080.
However, requests to the public domain still return:
404
x-railway-fallback: true
Server: railway-hikari
I also checked the deployment's Network Logs → HTTP immediately after accessing the public domain. The HTTP Network Logs show:
"No logs in this time range"
So the request is not reaching the service/container at all.
The application itself starts successfully and is ready to accept connections.
Could you please check the Railway edge/Hikari routing, domain-to-deployment binding, and replica routability/readiness on your side?
I have not modified PostgreSQL, run Prisma migrations, or changed the application source.
Attachments