Scheduling Component in an HA Cluster

In the production environment, multiple management nodes are deployed in a Kubernetes cluster to prevent the cluster from being unavailable when a single management node is faulty. Kubernetes provides two HA cluster setup solutions. For details, see "Options for Highly Available Topology" in the Kubernetes documentation. Cluster scheduling components are verified according to the official Stacked etcd topology solution, and each component can run properly when multiple management nodes are installed.

If multiple management nodes are deployed, ensure that their configurations are the same, such as the image of each cluster scheduling component, log directory, running user, and node label. For details about how to install cluster scheduling components on multiple management nodes, see Installation and Deployment. For details about how to install multiple management nodes, see "Creating Highly Available Clusters with kubeadm" in the official Kubernetes document.