Kube-vip for Control plane only - A lot of restarts

[tardich]

Hello.

I'm on K8S 1.29 (from a RKE2 cluster), and running kube-vip as control plane only HA, and seeing a lot of pod restarts that translate of multiple access loss with the API, which is kind of annoying.

The deployment is rather simple. Values are:

image:
  tag: v0.7.1
env:
  vip_interface: bond0
  vip_leaderelection: true
  svc_enable: false
  cp_enable: true
  vip_leaseduration: 5
  vip_renewdeadline: 3
  vip_retryperiod: 1
[...]

DaemonSet is running, obviously, on control nodes only. The pods look like:

kube-vip-g7d2z 1/1 Running 9 (2m26s ago) 66m
kube-vip-h5vqt 1/1 Running 6 (10m ago) 66m
kube-vip-nkhpr 1/1 Running 5 (2m13s ago) 66m

Logs showing:

level=fatal msg="lost leadership, restarting kube-vip"

Any clue?

[thebsdbox]

Is etcd/or the API server crashing a lot too? Also what is the infrastructure?