[rockythorn]

Summary

Add listen [::]:8086 to the nginx config so it accepts connections on both IPv4 and IPv6.

Problem

Pods in the apollo-frontend namespace are assigned IPv6 addresses (e.g., 2600:1f16:f2f:a006:2492::3). The Istio sidecar performs health checks against the pod's IP address, which means it probes nginx over IPv6. With only listen 8086 (IPv4), nginx doesn't bind to the IPv6 address, and the sidecar gets connection refused:

Request to probe app failed: Get "http://[2600:1f16:f2f:a006:2492::3]:8086/_/healthz":
  dial tcp [::6]:0->[2600:1f16:f2f:a006:2492::3]:8086: connect: connection refused,
  original URL path = /app-health/errata-frontend/readyz

This causes the pod to show 1/2 READY and eventually enter CrashLoopBackOff — the nginx container is healthy but the Istio sidecar reports the app as unreachable.

This is the same issue affecting the old apollo-frontend-deployment pods in the namespace.

Fix

listen 8086;
listen [::]:8086;

Nginx will now accept connections on both IPv4 and IPv6, allowing the Istio sidecar health checks to succeed.