I was working on my first cluster on VE 7.4.2 and trying to connect two nodes in a cluster. There are two nodes - A and B, where A is the one that owns the cluster, and B is attempting to join it. Note that Node B was cleanly installed and setup minutes before trying to join the cluster.
When I pass the join information to B it goes through the first few steps without failure, generating certificates and then randomly freezes. The web GUI becomes unable to load for both A and B, but I can ssh into A. To regain access I have to remove node B from the cluster.
After that there are no logs on Node A that indicate any sign of failure. Node B, however, says that Join Cluster failed and was unable to create /etc/pve/nodes because of a permissions issue. Note that I have done this both times where I have manually made that folder (using the same permissions as Node A, then same as everything in /etc/pve), with no success either time.
Another issue that I noticed was that Node A's corosync service was trying to retransmit a list each time. Node A has it's own NIC dedicated to the cluster, whereas Node B does not.
The only thread I could find that had similar information was for Proxmox VE 3.X and 2.X ... so those are not giving me any steps that I can run through.
When I pass the join information to B it goes through the first few steps without failure, generating certificates and then randomly freezes. The web GUI becomes unable to load for both A and B, but I can ssh into A. To regain access I have to remove node B from the cluster.
After that there are no logs on Node A that indicate any sign of failure. Node B, however, says that Join Cluster failed and was unable to create /etc/pve/nodes because of a permissions issue. Note that I have done this both times where I have manually made that folder (using the same permissions as Node A, then same as everything in /etc/pve), with no success either time.
Another issue that I noticed was that Node A's corosync service was trying to retransmit a list each time. Node A has it's own NIC dedicated to the cluster, whereas Node B does not.
The only thread I could find that had similar information was for Proxmox VE 3.X and 2.X ... so those are not giving me any steps that I can run through.
Last edited: