Users in Murmansk, Russia report that https://potok-kanban.vercel.app/ often fails to load with Chrome ERR_CONNECTION_RESET on both office Wi-Fi and mobile data without a VPN. The same site works via VPN and from outside Russia. The error occurs before the app or login screen appears.
Read-only tests on 2026-09-26 from a Russian Wi-Fi network forced curl to the physical en0 interface (bypassing the active VPN), with the same hostname and TLS SNI:
- 216.198.79.67: TCP connects; TLS times out (HTTP 000).
- 64.29.17.67: TCP connection times out.
- 216.198.79.195: HTTPS 200, valid certificate.
- 76.76.21.21: HTTPS 200, valid certificate. Earlier direct tests also timed out during TLS on 64.29.17.131 and 216.198.79.131, while VPN access worked. The Supabase backend was directly reachable from this network.
This suggests IP/path-dependent access, but we cannot tell whether the reset/timeout is caused by an upstream ISP/filter or Vercel ingress. The site uses a Vercel-managed .vercel.app domain; we do not own a custom domain and cannot change its DNS records. Could a Vercel networking specialist or community member help identify the cause and recommend a supported mitigation for the existing hostname? We are not asking to permanently pin traffic to 76.76.21.21.
No production code, deployment, or database configuration was changed during these tests.