Calico config breaks if you use etcd

[bsiagrac]

Environment:

• Cloud provider or hardware configuration:
  bare metal

• OS (printf "$(uname -srm)\n$(cat /etc/os-release)\n"):
  Ubuntu 20.04.6 LTS

• Version of Ansible (ansible --version):
  ansible [core 2.14.6]

• Version of Python (python --version):
  Python 3.8.10

Kubespray version (commit) (git rev-parse --short HEAD):
tag v2.23.0

Network plugin used:
calico with etcd

Description:
We encountered massive network problems, that if we use etcd in calico config, the configuration breaks during the upgrade to tag v2.23.0. The kubernetes internal network communication between the nodes was broken, after the upgrade. For example i/o timeouts and no route to host errors.

The calico-config configmap wrote the control-plane node name in the configmap and therefore all daemonsets wrote the same nodename in the config.

The fix was to replace the actual node name of the control plane with the variable __KUBERNETES_NODE_NAME__.

This was our fix

...
"plugins":[
        {
                                "nodename": "__KUBERNETES_NODE_NAME__",
                                "type": "calico",
            "log_level": "info",
                      "log_file_path": "/var/log/calico/cni/cni.log",
                                "etcd_endpoints": "https://[ETCD-IP]:2379",
            "etcd_cert_file": "/etc/calico/certs/cert.crt",
            "etcd_key_file": "/etc/calico/certs/key.pem",
            "etcd_ca_cert_file": "/etc/calico/certs/ca_cert.crt",
...

We assume that the error might be here roles/network_plugin/calico/templates/calico-config.yml.j2. If you use etcd you might need the same node configuration as if you use kdd but without the datastore variable.

This commit here might have broke the setting, when the config was changed to a configmap:
62f30a3

[VannTen]

duplicate of #10436
/close

[k8s-ci-robot]

@VannTen: Closing this issue.

In response to this:

duplicate of #10436
/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.