Kind: After docker-windows restart extraPortMappings not binding

[Haim-Bi]

What happened:
After install kind using kind create cluster --config config.yaml

apiVersion: kind.x-k8s.io/v1alpha4
kind: Cluster
networking:
  apiServerAddress: "127.0.0.1"
  apiServerPort: 6443
  podSubnet: "10.240.0.0/16"
  serviceSubnet: "10.0.0.0/16"
nodes:
- role: control-plane
  kubeadmConfigPatches:
  - |
    kind: InitConfiguration
    nodeRegistration:
      kubeletExtraArgs:
        node-labels: "ingress-ready=true"
  extraMounts:
  - hostPath: C:\local_deployment_files
    containerPath: /files
  extraPortMappings:
  - containerPort: 30000
    hostPort: 80
    listenAddress: "127.0.0.31"
    protocol: TCP
  - containerPort: 30011
    hostPort: 5001
    listenAddress: "127.0.0.31"
    protocol: TCP
  - containerPort: 30012
    hostPort: 5006
    listenAddress: "127.0.0.31"
    protocol: TCP

everything is working as expected.
once i restarted the KinD container or restarted docker-desktop
the ports are not bind anymore.

What you expected to happen:
The port will be bind as configured.

How to reproduce it (as minimally and precisely as possible):
Create new KinD cluster with extraPortMappings.
restart docker-desktop

Environment:
docker-desktop community 2.5.0.1

• kind version: (use kind version): kind v0.9.0 go1.15.2 windows/amd64
• Kubernetes version: (use kubectl version): v1.19.3
• Docker version: (use docker info): 19.03.13
• OS (e.g. from /etc/os-release): Windows 10 19041

[BenTheElder]

Can you debug more about what is and isn't working here?
Otherwise this just sounds like a bug in docker, kind is only responsible for telling docker to do the mapping, docker is responsible for actually binding ports 🤷‍♂️

[BenTheElder]

These are mapped to -p https://docs.docker.com/config/containers/container-networking/

This bug does not exist when using docker on other platforms as far as I can tell. I don't know that we can do anything about this here.

[Haim-Bi]

It looks like after the docker-desktop restart, the pods up time is still counting like there was no reset. Is KinD keeps it persist?

After redeploy the pods again everything is working.

[BenTheElder]

Kind uses persistent storage for pods yes.

[Haim-Bi]

Thanks.
Redeploy my deployment fixed that for me.
I'm closing this issue.

Thanks very much!

[BenTheElder]

I wonder if the deployment is stuck in some other way such as using cached DNS results for the node IPs?
It's still possible something is broken here, but it's not clear _where_.

[Haim-Bi]

Yes agreed. I will continue investigating this.
And update once i got something.

Thanks