3 days ago
Hello Railway support team,
My service now has two auto‑generated .up.railway.app domains:
- lisheng‑production.up.railway.app (the original domain I need working, currently bound to service)
- lisheng‑production‑15e0.up.railway.app (new generated domain for troubleshooting comparison, works perfectly both domestic and overseas).
Important note: The issue first happened when ONLY lisheng‑production.up.railway.app was attached to the service. The second domain was added later just for testing.
Both domains are set with explicit port forward to 8080.
Inside container echo $PORT returns 8080, my application listens correctly on 0.0.0.0:8080.
DNS resolution for lisheng‑production.up.railway.app works fine, it resolves to Railway edge IP 69.46.46.122.
But accessing this domain results in ERR_CONNECTION_TIMED_OUT, TCP connection times out before reaching HTTP layer.
What I have already tried:
I removed this domain from service, waited overnight for cache expiration, then re‑bound it back to my service.
The problem still persists after re‑binding.
This is stale/broken edge‑network state specific to this hostname. Please fully purge all edge network state for lisheng‑production.up.railway.app and restore connectivity.
Do NOT release this domain name.
Project ID: 761458a5‑a0a4‑4500‑9e83‑30458d790072
Service ID: 0cd34910‑f6a3‑4f73‑aa71‑059c01cc4f73
Thanks.
8 Replies
3 days ago
This thread has been opened as a bounty so the community can help solve it.
Status changed to Open Railway • 3 days ago
3 days ago
Have you tried using a VPN?
Also, I'd make sure the service is online and the URL is correctly typed. It's currently returning a 404.
3 days ago
I have tested with overseas VPN egress, and I still get ERR_CONNECTION_TIMED_OUT on lisheng‑production.up.railway.app.
The 404 you observed only occurs when hitting a small subset of healthy edge PoPs. Many other global edge nodes for this exact hostname result in full TCP handshake timeout before traffic can reach my container.
For comparison:
lisheng‑production1.up.railway.app runs on the exact same service with identical port‑forward settings and works reliably across every edge node worldwide, no timeouts at all.
I already removed this domain, waited overnight and re‑bound it to my service, and this partial‑PoP routing bug remains.
Please do not treat your single 404 test result as full service health. This hostname has stale broken state spread across multiple global edge locations, and requires a full purge across ALL PoPs. Please do NOT release this domain name.
Project ID: 761458a5‑a0a4‑4500‑9e83‑30458d790072
Service ID: 0cd34910‑f6a3‑4f73‑aa71‑059c01cc4f73
0x5b62656e5d
Have you tried using a VPN? ~~Also, I'd make sure the service is online and the URL is correctly typed. It's currently returning a 404.~~
3 days ago
I have tested with overseas VPN egress, and I still get ERR_CONNECTION_TIMED_OUT on lisheng‑production.up.railway.app.
The 404 you observed only occurs when hitting a small subset of healthy edge PoPs. Many other global edge nodes for this exact hostname result in full TCP handshake timeout before traffic can reach my container.
For comparison:
lisheng‑production1.up.railway.app runs on the exact same service with identical port‑forward settings and works reliably across every edge node worldwide, no timeouts at all.
I already removed this domain, waited overnight and re‑bound it to my service, and this partial‑PoP routing bug remains.
Please do not treat your single 404 test result as full service health. This hostname has stale broken state spread across multiple global edge locations, and requires a full purge across ALL PoPs. Please do NOT release this domain name.
Project ID: 761458a5‑a0a4‑4500‑9e83‑30458d790072
Service ID: 0cd34910‑f6a3‑4f73‑aa71‑059c01cc4f73
3 days ago
Correction: I'm able to access lisheng-production.up.railway.app just fine from Toronto, Canada, and through ProtonVPN (Japan). I'd try using a different VPN.
Attachments
3 days ago
Critical requirement: this website must work reliably in mainland China, as many competition judges are based in China.
Right now:
- Canada VPN: lisheng-production.up.railway.app works fine (404 response, Canada edge PoP healthy).
- Mainland China network / VPN regions outside Canada: same hostname hits ERR_CONNECTION_TIMED_OUT, TCP handshake fails.
This is partial stale routing state on Cloudflare edge PoPs, specifically the PoPs serving China/Asia-Pacific traffic.
The domain is locked for competition submission, I cannot change the submitted URL.
The reference service lisheng-production1.up.railway.app runs on identical deployment and works perfectly from mainland China and all global regions.
URGENT: please fully purge lisheng-production.up.railway.app across ALL Cloudflare edge PoPs, especially the China/Asia-Pacific edge locations.
Do NOT release this domain name.
Project ID: 761458a5-a0a4-4500-9e83-30458d790072
Service ID: 0cd34910-f6a3-4f73-aa71-059c01cc4f73
3 days ago
lisheng-production.up.railway.app hits Railway's edge, not Cloudflare's edge. If users in China aren't able to access the site, their only option would be to use a VPN. There isn't anything Railway can do about China's restrictions.
0x5b62656e5d
`lisheng-production.up.railway.app` hits Railway's edge, not Cloudflare's edge. If users in China aren't able to access the site, their only option would be to use a VPN. There isn't anything Railway can do about China's restrictions.
3 days ago
But lisheng-production1.up.railway.app works perfectly from mainland China.