Installation Environment Setup

When Kubernetes uses Calico as the cluster network plugin, the node-to-node mesh network configuration is used by default. For large-scale clusters, this configuration may cause heavy load on the service switch network, so you are advised to use the reflector mode. For details, see Calico official documentation.
If Kubernetes is installed on CentOS 7.6 and Calico v3.24 is used as the cluster network plugin, the operation may fail. For details, see System requirements.
In Kubernetes 1.24 and later versions, dockershim has been removed from Kubernetes projects. If you want to use Docker as the container engine of Kubernetes, you need to install cri-dockerd. For details, see What Can I Do If Docker Fails to Be Used in Kubernetes 1.24 or Later?.
Kubernetes 1.25.10 and later versions do not support vNPU recovery.

Before installing cluster scheduling components, ensure that the following software has been installed:

Docker (any version in 18.09.x to 28.5.1). For details, see Install Docker Engine.
containerd (any version in 1.4.x to 2.1.4). For details, see Installing containerd.
Kubernetes (any version in 1.17.x to 1.34.x; version 1.19.x or later is recommended). For details, see Installing Kubernetes with deployment tools. Creating a cluster with kubeadm is recommended. Rectify faults during cluster initialization by referring to Kubernetes Initialization Failed. In addition, de-isolate the management node. The following commands are for reference only.
- De-isolate a single node:
```
kubectl taint nodes <hostname> node-role.kubernetes.io/master-
```
- De-isolate all nodes:
```
kubectl taint nodes --all node-role.kubernetes.io/master-
```
  By de-isolating the management node, the taint on the primary node is removed. This allows pods to be scheduled to the primary node.

Parent topic: Installation