Scenario Description

When a node in the Proxmox VE cluster is damaged and cannot be quickly repaired, it is necessary to cleanly remove the faulty node from the cluster and clear out any residual information. Otherwise, a new node using the same IP address as the faulty node will not be able to join the cluster properly. Similarly, if the faulty node is repaired after being removed from the cluster, even though it has no relation to the cluster, accessing the web management interface of this single node will display information about other nodes from the original Proxmox VE cluster, which can be very frustrating.

Removing a Node from the Cluster

If the Proxmox VE is part of a Ceph hyper-converged cluster, log into any node in the cluster (except the one to be deleted) on the host system Debian, and thoroughly delete all OSDs of the node to be removed using the command line. Continue using the command “pvecm delnode nodename” to remove the node. Switch to the Proxmox VE cluster web management interface, refresh the browser, and check if the expelled node still exists on the left side of the “Server View” page. Even if the web page no longer shows the expelled node’s information, further steps are necessary to completely clean up the expelled node’s information.

Switch to the host system Debian of any node in the Proxmox VE cluster

In the command line, navigate to the directory “/etc/pve/priv” and edit the files “authorized_keys” and “known_hosts”.

Use the Vi editor or Sed command to delete the information of the node expelled from the cluster, save and exit. The other nodes in the cluster will automatically synchronize via corosync, so no further action is needed.

Similarly, edit the “Known_hosts” file. After completing this step, use the command “grep pve169 /etc -r” to search and ensure complete removal.

Handling Residual Information from the Removed Node

If the node removed from the Proxmox VE cluster is restored and intended to rejoin the original cluster without reinitialization (reinstalling the system), you may still see information about other nodes in the cluster when accessing the Proxmox VE web management interface of this single node, even after refreshing the page or rebooting the system. So, how can this annoying display be eliminated?

In addition to stopping the cluster service, stopping the Corosync synchronization service, and deleting corosync.conf, you also need to log into the Proxmox VE single node host system Debian, navigate to the directory “/etc/pve/nodes”, and delete the directories of all nodes except for the current one.

After deletion, refresh the browser, and the Proxmox VE single node web management interface will only show the single node itself.

After completing the above two steps, the node can rejoin the cluster with the original IP without being rejected by the cluster.