2 months ago
Hi Railway team — we have an orphaned generated hostname still serving traffic after the service was deleted.
Project: lode-automation
Project ID: de0fd422-1521-4573-bae1-18a934babe63
Environment ID: 3c72d021-b0ef-4e2c-a07d-e7932af2e657
Stale hostname:
https://openclaw-gateway-production.up.railway.app
The service openclaw-gateway and its volume have already been deleted, but this hostname still responds with HTTP/2 200 and serves HTML from railway-edge.
Evidence:
curl -I https://openclaw-gateway-production.up.railway.app/
curl -I https://openclaw-gateway-production.up.railway.app/healthz
curl -I https://openclaw-gateway-production.up.railway.app/readyz
All return HTTP/2 200.
The new guarded hostname works correctly:
https://openclaw-gateway-production-0c3f.up.railway.app
(returns 404 at /)
This stale hostname appears to be an orphaned edge route that still points to a deleted service.
Could you please remove or detach the route for
openclaw-gateway-production.up.railway.app?
This is currently a stop-ship security blocker for us because it exposes an admin/control surface publicly.
Thanks!
1 Replies
Status changed to Awaiting Railway Response Railway • 2 months ago
2 months ago
Hey richext, that definitely sounds like a stale edge cache/routing table issue where the control plane's 'delete' event didn't propagate to the edge nodes.
Since this is a security blocker, you might be able to 'force-sync' it yourself while waiting for the team:
- Create a 'Shadow' Service: Create a new Empty Service in the same environment and name it exactly
openclaw-gateway. - Reclaim the Domain: Go to the Networking tab of this new service. Try to manually add the custom/generated domain
openclaw-gateway-production.up.railway.app. - The 'Null-Route' Fix: If it lets you claim it, the edge should now point to your new (empty) service instead of the ghost one. You can then either leave it as a 404 or delete the domain link manually before deleting the service again. This usually forces the edge proxy to overwrite the stale mapping.
Technical Note for Railway: It looks like the Edge routing table is out of sync with the Service Discovery layer for Environment 3c72d021-b0ef-4e2c-a07d-e7932af2e657. The /healthz and /readyz returning 200s from a deleted service suggests the edge might still be holding onto a persistent connection or a stale container IP mapping.