Ubuntu Install Kubernetes with Kubeadm Issues

Due to the requirements of the new versions of Docker and Kubernetes for the system, the following operations must first be performed to ensure that Docker and kubelet can work properly.

First, modify or add the /etc/docker/daemon.json file with the following content:

{
     "registry-mirrors": ["https://94zlnekp1.mirror.aliyuncs.com"],
     "log-driver": "json-file",
     "log-opts": {
       "max-size": "10m",
       "max-file": "1"
      },
     "exec-opts": ["native.cgroupdriver=systemd"]
}

94zlnekp1.mirror.aliyuncs.com is a personal Docker acceleration address.

Restart Docker:

sudo systemctl daemon-reload  
sudo systemctl restart docker  

This step primarily addresses the cgroups issue.

This error is likely caused by:
	- The kubelet is not running
	- The kubelet is unhealthy due to a misconfiguration of the node in some way (required cgroups disabled)

Next, disable swap, which is a requirement for kubelet.

Turn off virtual memory:

swapoff -a  

Modify the configuration:

nano /etc/fstab  

Delete this line:

/swap.img	none	swap	sw	0	0  

Then delete the swap.img file.

Next, restart the container runtime:

systemctl restart containerd  

If you check the logs with journalctl -xeu kubelet and find:

"Error getting node" err="node \"k8smain\" not found"

You can open the /etc/hosts file and add or modify the IP corresponding to the current machine name.

127.0.0.1  myhost  

If multiple nodes are needed, then each node must be configured with the internal IP of the master node.

If you encounter:

r="failed to run Kubelet: unable to determine runtime API version: rpc error: code = Unavailable d

Then run:

crictl config runtime-endpoint unix:///run/containerd/containerd.sock  
crictl config image-endpoint unix:///run/containerd/containerd.sock  

Then check:

systemctl status containerd -l