"FailedToStartServiceHealthcheck... bind: address already in use" on dual-stack cluster

[h3mmy]

Details

NOTE: This is not a new issue. I've been waiting on an upstream fix. Just want to document it for my future self.
What steps did you take and what happened:

Check events. Routinely see events like this one:

53m         Warning   FailedToStartServiceHealthcheck   service/lldap-ldaps     node slowpoke failed to start healthcheck "auth/lldap-ldaps" on port 32056: listen tcp 0.0.0.0:32056: bind: address already in use

What did you expect to happen:

No FailedToStartServiceHealthcheck events

Anything else you would like to add:

Additional Information:

This is been there for years, but it is annoying. The root cause is the way kube-proxy handles dual-stack networks and is already reported in the kubernetes project at kubernetes/kubernetes#114702

For ref a sudo netstat -tulnp | grep LISTEN shows this for port 32056

tcp6       0      0 :::32056                :::*                    LISTEN      873/k3s server