We're experiencing intermittent getaddrinfo ENOTFOUND errors when calling our service web-production-4fac9.up.railway.app from an external production server (not a local machine — reproduced via direct curl from our partner's production server, ruling out client-side network issues).

Pattern observed: the first request to the domain fails with ENOTFOUND; subsequent requests within seconds succeed. This has been reproduced consistently across multiple sessions over two days.

What we've already ruled out on our end:

• sleepApplication: false — service does not idle/sleep.
• No 502/503/522/524 in our Railway logs during the failure windows (expected, since DNS resolution failures never reach our edge/server).
• No IPv6/dual-stack ambiguity — only an A record exists for our domain.
• Confirmed via nslookup/dig that the domain resolves correctly most of the time; the failure is intermittent, not persistent.

Additional finding: we also observed socket.gaierror: Temporary failure in name resolution while using railway ssh to connect to our own Postgres instance, from within Railway's own infrastructure — suggesting this may not be isolated to external callers.

We found similar reports in your community forum matching this exact symptom:

• https://station.railway.com/questions/intermittent-network-dns-errors-4c139d9d
• https://station.railway.com/questions/api-intermittently-unavailable-error-3ab1226f

Could you investigate whether there's a known issue with DNS resolution on the shared *.up.railway.app edge, particularly around the public networking layer (we noticed your status history shows 99.42% for "public networking" in some US regions during July)?

Service: web-production-4fac9.up.railway.app (also reachable via custom domain api.naap.nouxtech.com, same underlying infra).