Bug report

General Information

• Cilium version 1.8.3
• Kernel version 5.8.7
• Kubernetes 1.19.1
• cilium-sysdump-20200914-131007.zip

How to reproduce the issue

1. Install K8s 1.19.1 using kubeadm with IPv6 only, skipping addon/kube-proxy
2. Install cilium 1.8.3 using the attached config
3. Monitor logs and READY state of "cilium" pod

• issue: cilium never becomes READY, health endpoint restarting endlessly

4. Create a pod like this and try to "ping6" your default gateway by IP (IPv6)

• issue: connections or pings to nodes outside the cluster not possible, even if IP ranges are fully routable