Since the .up.railway.app endpoint responds in ~1 second but the custom domain takes 15–20 seconds, it suggests the application itself is probably not the bottleneck. I'd focus on the path that's unique to the custom domain.

A few things to check:

1. Measure where the delay occurs using:

   
   curl -v https://api.rutadigitalcrm.com
   

   or

   
   curl -w "DNS: %{time_namelookup}\nTCP: %{time_connect}\nTLS: %{time_appconnect}\nTTFB: %{time_starttransfer}\nTotal: %{time_total}\n" -o /dev/null -s https://api.rutadigitalcrm.com
   

   This will show whether the delay is in DNS, TCP, TLS, or waiting for the first byte.

2. Compare the results with the .up.railway.app URL to identify which stage differs.

3. If you're using Cloudflare or another reverse proxy, temporarily disable the proxy (DNS-only) and test again. If the latency disappears, the issue is likely in the proxy path rather than Railway.

4. Test IPv4 vs IPv6 separately:

   
   curl -4 https://api.rutadigitalcrm.com
   
   curl -6 https://api.rutadigitalcrm.com
   

   A large difference may indicate an IPv6 routing issue.

5. Verify the custom domain resolves only to the expected Railway endpoint and that there aren't stale or duplicate DNS records.

Knowing which phase (DNS, TCP, TLS, or TTFB) accounts for the extra 15–20 seconds will make it much easier to identify the root cause.