Hello Vercel Team,
I am experiencing intermittent connectivity issues with my production website hosted on Vercel.
Project Information • Domain: https://moshal.vercel.app • Framework: Next.js 16 • Hosting: Vercel Hobby • Location: Bangladesh • Tested networks: Robi and Airtel/Circle mobile data
Problem Description
The website intermittently becomes unreachable and the browser shows connection timeout errors.
The issue sometimes resolves automatically after several minutes without any redeployment or code changes.
I investigated the issue using curl, DNS lookup, MTR, and the official Vercel connectivity diagnostic script.
IP Connectivity Test Results
I tested eight IP addresses associated with my domain.
| IP Address | Result |
|---|---|
| 216.198.79.3 | TCP timeout |
| 216.198.79.67 | HTTP 200 |
| 216.198.79.131 | TCP timeout |
| 216.198.79.195 | TCP timeout |
| 64.29.17.3 | TCP timeout |
| 64.29.17.67 | HTTP 200 |
| 64.29.17.131 | TCP timeout |
| 64.29.17.195 | HTTP 200 |
Important Observations
- The same domain successfully returns HTTP/2 200 when accessed through certain IP addresses.
- Other IP addresses consistently fail during TCP connection establishment on port 443.
- The failures occur before TLS handshake or HTTP communication.
- DNS resolution returns IP addresses that are not reachable from my tested networks.
- Similar IP specific connectivity failures were observed on both Robi and Airtel/Circle mobile networks.
- The issue occurs without any deployment or application code changes.
Diagnostic Evidence
I have collected: • Official Vercel connectivity diagnostic output. • DNS lookup results using the local resolver and Google DNS. • MTR reports for working and failing IP addresses. • Detailed curl connection logs. • Comparative tests across two mobile network operators.
What I Need Help With
Could the Vercel team investigate:
- TCP/443 reachability to the affected Vercel edge IP addresses from Bangladesh.
- Possible regional edge connectivity, BGP routing, or upstream peering issues.
- Why some IP addresses successfully serve the same production deployment while others time out.
- Whether there is any recommended mitigation or supported configuration for this issue.
This is affecting the availability of my production website, so I would appreciate help identifying the underlying cause.
I have the complete vercel-debug.txt diagnostic report available and can share it if required.
I can provide the complete diagnostic output and additional test results if needed.
Thank you.