an hour ago
Since at least 2026-11-04, our VPS at 201.18.192.x (Hostinger, AS47583, Mumbai) cannot reach any Railway edge IP. dlapi-production.up.railway.app (69.46.46.115), another of our services (69.46.46.92) and railway.com (69.46.46.46) all show 100% loss on ICMP, TCP/80 and TCP/443. A packet capture shows our SYNs leaving and no SYN-ACK, RST or ICMP coming back.
The path goes Hostinger → Airtel (AS9498) → CDN77 Singapore (airtel-sgp.cdn77.com 138.199.1.230, then vl222.sgp-m1-dist-1.cdn77.com 79.127.194.51). It dies at the next hop, i.e. at the handoff into Railway's Singapore edge. Those CDN77 hops also show up to 3 s latency and up to 70% loss. Other destinations from the same host work normally (Google: HTTP 200 in 0.35 s).
The service itself is healthy and reachable from other networks; the problem is only this ingress path. Full traces are attached (source IP masked; we can share it privately if needed). Could you check the Singapore edge's peering with CDN77/Airtel?
1 Replies
an hour ago
Our edge addresses in 69.46.46.0/24 are a shared anycast pool. We can't change filtering or loss on networks outside ours, and we can't assign or exclude specific edge IPs for one service. Redeploying or changing region also won't change the edge IP your traffic lands on.
If that upstream path keeps the VPS from reaching the edge, the lasting option is to put a proxy or CDN in front of the domain, one whose IPs that host can reach.
Your traces from the VPS to 69.46.46.46 are the data your hosting provider or Airtel can act on. Running the Network Diagnostics tool from an affected machine gives them more detail on the path.
Status changed to Awaiting User Response Railway • about 1 hour ago