Hi Vercel team — I’m seeing what looks like a custom-domain edge/TLS/routing issue on a Hobby project.
Summary
- My project’s native
*.vercel.appdeployment works normally - The custom apex domain and
wwwsubdomain are both attached to the correct project - In the Vercel dashboard, both show:
- Valid Configuration
- Production
- But the branded/custom domain path is inconsistent depending on the edge path
What I verified
- The Vercel deployment itself is healthy because the
*.vercel.apphostname works - The custom domains are bound to the right project in Vercel
- DNS is pointing to Vercel targets using external DNS
- I already tried the recommended workaround:
- removed both custom domains
- waited
- re-added them
- confirmed they reassigned to production
- The issue still persists after re-adding
Debug findings
I ran the Vercel connectivity debug script, and the results were inconsistent across the resolved Vercel edge IPs for the custom domain:
-
one edge IP:
-
completes TLS successfully
-
presents the correct certificate for the custom domain
-
but then the HTTP request hangs / times out with 0 bytes returned
-
the other tested Vercel edge IPs:
-
reset during TLS handshake (
connection reset by peer)
So this does not look like:
- an app/runtime problem
- a simple DNS typo
- or a stale local DNS cache issue
It looks more like:
- inconsistent custom-domain edge behavior
- or an issue in TLS/routing/certificate state across edge nodes
Why I’m posting
Because:
*.vercel.appworks- custom domain config shows valid
- remove/re-add did not fix it
- edge behavior appears inconsistent across Vercel-served IPs
Has anyone seen this before, or can a Vercel team member help review the edge/certificate/routing state for a custom domain showing this pattern?
I have the full vercel-debug.txt output and can share it privately with Vercel staff if needed.
If a team member wants the full debug output, please reply here and I can provide it directly.