Confirm IPv6 binding issue causing 502 Bad Gateway in Cased CD

confirm-ipv6-issue.md

Confirm IPv6 Binding Issue and Test Fix

The Problem

The backend is listening on IPv6 only (:::8081), but nginx is trying to connect via IPv4.

Confirmation Tests

1. Confirm IPv6-only binding

# Check if backend responds on IPv4 loopback
curl -v --ipv4 http://127.0.0.1:8081/health

Expected: Connection refused or timeout

# Check if backend responds on IPv6 loopback
curl -v --ipv6 http://[::1]:8081/health

Expected: Should return "OK" ✅

2. Confirm nginx can't reach it

# Test what nginx is doing
curl -v http://localhost:8080/api/v1/settings/audit

Expected: 502 Bad Gateway (current issue)

Quick Fix Test

Try forcing the backend to listen on IPv4:

Option A: Test with netcat (if available)

# Kill the current backend (it will restart)
pkill cased-backend

# Wait for restart, then check binding again
sleep 3
netstat -tlnp | grep 8081

Option B: Test if ::1 works from nginx

# Check if we can reach backend via IPv6 localhost
curl http://[::1]:8081/health

Expected: Should return "OK"

The Real Fix (if confirmed)

If IPv6 works ([::1]:8081) but IPv4 doesn't (127.0.0.1:8081), then we need to:

Either:

1. Make backend bind to 0.0.0.0:8081 (IPv4) explicitly
2. Make nginx use [::1]:8081 instead of localhost:8081
3. Enable IPv6 in nginx resolver config

Quickest Workaround (For Testing)

If you want to test if this fixes it immediately:

# Inside the pod, try accessing via IPv6 through nginx
# (This won't work directly, but confirms the issue)

# Check if curl can reach via IPv4 vs IPv6
curl --ipv4 -I http://localhost:8081/health
curl --ipv6 -I http://localhost:8081/health

If IPv6 works and IPv4 doesn't, that's the smoking gun 🔫

Expected Results

• ❌ curl --ipv4 http://127.0.0.1:8081/health fails
• ✅ curl --ipv6 http://[::1]:8081/health works
• ❌ curl http://localhost:8080/api/v1/settings/audit fails (502)

If all three match expectations above, the fix is confirmed and we should change the Go backend to bind explicitly to IPv4.