Railway production outbound TCP connection times out to bseindices.com:443

I’m troubleshooting a production networking issue on Railway.

My service can resolve:

bseindices.com -> 43.228.177.148

However, repeated outbound TCP connections from the Railway production service to:

43.228.177.148:443

do not complete and eventually time out.

Application diagnostics repeatedly show:

DNS_OK | host=bseindices.com | ips=('43.228.177.148',)

followed by:

TCP_TLS_FAILED | ip=43.228.177.148 | error_type=TimeoutError | error=timed out

Railway network-flow captures repeatedly show outbound TCP flows to 43.228.177.148:443 with:

• direction=egress
• l4Protocol=tcp
• flowState=partial
• dropCause=null
• l4LatencyMs=0.0
• packet counts between 1 and 5
• no completed TCP flow observed in the captured records

The same endpoint was independently tested from a GitHub Codespace and works normally over IPv4:

• TCP connection succeeds
• TLS 1.3 succeeds
• HTTP 200 OK

Measured control request:

HTTP=200 CONNECT=0.005667 TLS=0.032739 TOTAL=0.057378

This appears to be a connectivity-path difference between the Railway production environment and the external endpoint, rather than an HTTP/application parsing problem.

The application is intentionally fail-closed when this authoritative reference source is unavailable, so no trading bypass or workaround is being requested.

Could anyone help determine whether this pattern indicates an outbound routing/NAT/egress issue, an upstream connectivity issue, or another Railway networking condition?

In particular, what does a repeated partial TCP flow with dropCause=null and no completed connection indicate, and what Railway-side diagnostic should be used next?

I can provide deployment/service identifiers and additional logs in the thread if needed.