MTN Rwanda cannot reach Vercel edge IPs 64.29.17.x / 216.198.79.x, while legacy Vercel IPs work

MTN Rwanda cannot reach Vercel edge IPs, while legacy Vercel IPs work

Hello Vercel team,

I am experiencing a connectivity issue with my production Vercel deployment:

https://rudakemwa-portal.vercel.app/

rudakemwa-portal.vercel.app

Rudakemwa Agribusiness Portal

The deployment itself appears healthy, but the site cannot be reached from MTN Rwanda without a VPN/WARP connection.

Environment

ISP: MTN Rwandacell
ASN: AS36890
Location: Kigali, Rwanda
Public IP during testing: 41.186.133.43

What I found

The Vercel hostname resolves to Vercel's newer edge IP ranges:

216.198.79.0/24
64.29.17.0/24

From MTN Rwanda, TCP connections to these Vercel edge IPs on port 443 consistently time out.

I tested all of these addresses:

216.198.79.3
216.198.79.67
216.198.79.131
216.198.79.195
64.29.17.3
64.29.17.67
64.29.17.131
64.29.17.195

All timed out on TCP/443.

However, when testing the same deployment against legacy Vercel IPs:

76.76.21.21
66.33.60.130

the connection succeeds and returns:

HTTP/1.1 200 OK Server: Vercel X-Vercel-Cache: HIT

The Vercel website itself is also reachable from the same MTN connection.

Additional evidence

The deployment works immediately when using WARP/VPN.
The problem occurs from both my PC and phone on the normal MTN connection.
Traceroutes to the affected Vercel edge IPs reach MTN/upstream network hops and then time out.
Using different DNS resolvers changes which Vercel edge IP is returned, but the affected Vercel edge ranges remain unreachable.
Direct HTTPS testing confirms that the newer edge IPs time out before TLS, while the legacy Vercel IPs successfully return HTTP 200.

This strongly suggests a routing, Anycast, BGP, peering, or upstream connectivity issue between MTN Rwanda and Vercel's newer edge IP ranges rather than an application, backend, CORS, or DNS configuration issue.

Could the Vercel Network/Infrastructure team please investigate the routing/connectivity to these edge ranges from AS36890 (MTN Rwanda)?

I can provide the full Vercel Network Debug output, traceroutes, curl results, and timestamps if needed.

Thank you.

The domain troubleshooting guide can help with most custom domain configuration issues. You might be able to use that guide to solve it before a human is available to help you. Then you can come back here and share the answer for bonus points.

https://vercel.com/docs/domains/troubleshooting

vercel.com

Troubleshooting domains

Learn about common reasons for domain misconfigurations and how to troubleshoot your domain on Vercel.

Additional evidence:

The deployment works immediately when I use WARP/VPN.
The same issue occurs from both my PC and phone on the normal MTN Rwanda connection.
vercel.com itself is reachable from the same MTN connection.
Traceroutes to the affected Vercel edge IPs reach MTN/upstream network hops and then time out.
Direct HTTPS testing times out before TLS on the newer edge IPs.
The legacy Vercel IPs 76.76.21.21 and 66.33.60.130 successfully return HTTP 200 for the same deployment.

This appears to be a network routing / Anycast / BGP / peering issue affecting connectivity from MTN Rwanda (AS36890) to Vercel's newer edge ranges.

Could someone from Vercel Network/Infrastructure investigate the connectivity from AS36890 to 64.29.17.0/24 and 216.198.79.0/24?

I can provide the full Vercel Network Debug output and traceroutes if required.

I’m seeing the same failure from T-Mobile 5G Business Internet in the United States.

Affected deployment: https://five-circa-dashboard.vercel.app/

five-circa-dashboard.vercel.app

Circa HQ / 2026

Circa HQ: the team's Circa contest scorecard.

Evidence from the affected connection:

DNS resolves normally to Vercel edge addresses in 64.29.17.0/24 and 216.198.79.0/24.
TCP/443 fails before TLS. curl reports connection failure and the browser reports ERR_ADDRESS_UNREACHABLE.
ping and traceroute to 64.29.17.3 return ‘No route to host’ immediately.
An unrelated vercel.app deployment and vercel.com are also unreachable from the same connection.
Google and other non-Vercel destinations remain reachable.
The deployment loads immediately after switching the same Mac to a phone hotspot.

This appears to be an outbound routing/peering issue between T-Mobile Business Internet and Vercel’s newer 64.29.17.0/24 and 216.198.79.0/24 edge ranges, rather than a deployment or DNS problem. Could Vercel Network/Infrastructure investigate and coordinate with T-Mobile? I can provide further traceroute or network-debug output if useful.