Thanks for the detailed report, that header and traceroute info is exactly what we need.

Quick clarification on how this works: your deployment region (Singapore, asia-southeast1-eqsg3a) controls where your container runs, but the edge POP you enter through is selected by anycast. Which edge you land on is decided by BGP, that is your ISP's and the upstream transit/CDN77 path, not by your deployment region. So it is expected for those two to be independent, and a Singapore deployment can still be reached via a non-Asia edge if your ISP is announcing a shorter path to that POP.

What you are seeing (Indonesia entering through lax1 instead of an Asia edge, with latency jumping from ~25ms to ~196ms day-over-day) points to an upstream routing/peering change at your ISP or in CDN77's transit, not a change on your service. Your deployment is healthy and the Singapore edge is operational on our side.

To dig in, can you send:

• A full traceroute/mtr from the affected network to your service hostname (both home ISP and mobile)
• Your ISP / ASN and rough city
• The exact times the latency changed

We will pass the LAX-from-Indonesia anycast path to CDN77 to check their peering. In the meantime this is a transit-side route selection, so it should recover as the upstream path settles.

— Angelo

City: Semarang, Central Java, Indonesia

Home ISP: IndiHome / Telkom Indonesia

I can provide a traceroute from my home IndiHome connection now. I can collect a mobile data traceroute tomorrow

Home ISP traceroute:

Tracing route ...

over a maximum of 30 hops:

1 4 ms 2 ms 1 ms 192.168.68.1

2 5 ms 1 ms 1 ms 192.168.1.1

3 12 ms 7 ms 4 ms 10.96.0.1

4 2 ms 2 ms 3 ms 180.252.4.97

5 * 21 ms 21 ms 180.240.190.61

6 22 ms 21 ms * 180.240.190.61

7 50 ms 24 ms 33 ms 180.240.190.230

8 195 ms 194 ms 193 ms 180.240.192.154

9 197 ms 196 ms 196 ms 101.203.74.30

10 196 ms 196 ms 195 ms vl221.lax-cyx1-dist-2.cdn77.com [79.127.195.164]

11 197 ms 195 ms 195 ms 69.46.46.27

Trace complete.

Mobile network also shows x-railway-edge: lax1 and approximately the same latency (~196 ms).

I don't have the exact timestamp for the routing change.

On 22 Jun 2026 I was consistently seeing approximately 25 ms latency.

On 23 Jun 2026 around 10:00–11:00 WIB I noticed latency had increased to approximately 196 ms and x-railway-edge was reporting lax1.

Thanks, this is enough detail to pass along: Semarang / Telkom Indonesia, latency change noticed June 23 around 10:00-11:00 WIB, and the route entering CDN77 via LAX.

We will pass this path to the edge/provider side for review. A mobile traceroute is still useful if you can capture it, but the home ISP trace already shows the important part: traffic from Indonesia is being handed to lax1 rather than an Asia POP.

The deployment itself is still in Singapore; this is about anycast/peering path selection into the Railway edge.