Cluster not ready - no quorum? (500)

To resolve the issue of the cluster not being ready and the “no quorum” or “Cannot initialize CMAP service” errors, you can follow the steps below:

  1. Stop the corosync and pve-cluster services on the node:

    systemctl stop pve-cluster
    systemctl stop corosync
    
  2. Start the cluster filesystem in local mode:

    pmxcfs -l
    
  3. Delete the corosync configuration files:

    rm /etc/pve/corosync.conf
    rm /etc/corosync/*
    
  4. Start the filesystem as a normal service:

    killall pmxcfs
    systemctl start pve-cluster
    
  5. Remove the separated node from the remaining node in the cluster using the pvecm delnode command. Replace oldnode with the name of the node you want to remove:

    pvecm delnode oldnode
    

    If the command fails because the remaining node lost quorum when the separated node exited, you can set the expected votes to 1 as a workaround:

    pvecm expected 1
    

    Then repeat the pvecm delnode command.

  6. Switch back to the separated node and delete any remaining files from the old cluster:

    rm /var/lib/corosync/*
    
  7. If there are configuration files from other nodes in the cluster filesystem, you may want to clean those up as well. Delete the entire directory recursively from /etc/pve/nodes/NODENAME, where NODENAME is the name of the node you separated. Make sure to double-check and confirm the correct directory before deleting it.

  8. Remove the SSH keys of the nodes from the authorized_keys file to prevent further connections:

    vi /etc/pve/priv/authorized_keys
    

    Inside the file, remove the respective public keys of the nodes that were separated.

By following these steps, you should be able to separate the node from the cluster and ensure clean separation of shared resources.