Issue 1
In the high-availability RKE2 Kubernetes cluster, I have provisioned a single load balancer (1LB) and three master nodes. Initially, all components were functioning as expected. However, following a reboot of one of the nodes, it failed to reestablish connectivity with the existing Kubernetes cluster. Upon executing systemctl restart rke2-server, the node started as a new server with only one node.
In a High Availability (HA) RKE2 cluster configuration with three nodes, the failure of two nodes results in the unavailability of the etcd server. Upon inspection, both the API server and the ETCD server are reported as unavailable. Consequently, I am unable to manage the Kubernetes cluster. This situation has resulted in data loss for the applications running on the cluster. Despite my restoration efforts, I have been unable to regain access to the RKE2 server running Kubernetes version 1.27.12.
Issue 2
Recently, I modified the hostname of a Linux server from its default value (VMI292921) to ‘rke2-master01.’ I subsequently configured it to be part of a Kubernetes (K8s) cluster. However, during reboot operations, the hostname reverts to the default name, causing issues with the K8s node joining the existing cluster.
After experiencing the failure of two nodes, I proceeded to insert a new node using the same configuration file. However, despite this action, I am still unable to access the Kubernetes cluster. The error message persists, indicating that the API server is not yet available and that the kube-proxy configuration retrieval is pending. My objective remains to regain access to the cluster even after the nodes have encountered issues
config.yaml file
server: https://load-balancer:9345
token: random token key with k10
tls-san:
- load-balancer-ip
