I used this repo as a starting point for K8s deployment and also got stuck at the clustering problem of confluence-dc.
I solved that and successfully run Confluence on Kubernetes by generating n services (n=replicas), each with:
type: ClusterIP
ClusterIP: 10.10.10.{{254-n}}
selector:
app: confluence-{{n}}
this way, you have static IPs to put into the config to use tcp_ip for cluster.type.
That method was more K8s-y than using VPN-sidecars for static IPs or such.
Just a suggestion to give back to you, thanks for helping me with your code :)