Split cluster after upgrade

[pjotter]

Hi,

After upgrading a four node 4.4 cluster to 4.4-20, one node restarts disconnected. No matter what we try, it cannot see the cluster, nor can the cluster see the node. All network interfaces are up and running. Network traffic is possible as before, we can see no reason why this node should not work. The linux containers it contains cannot restart, as there is no quorum. Reinstalling involves migrating the containers off the node, which is blocked by its disconnected state. Besides, we doubt reinstalling will solve the problem.

Is there any experience with this? Tips? Tricks?

 

[dietmar]

Please check if multicast works (omping).

https://pve.proxmox.com/wiki/Multicast_notes

 

[pjotter]

Multicast is disabled. The cluster is working with unicast.

 

[pjotter]

Please note that three nodes are running in perfect order.

 

[pjotter]

Ok, as this seems to have no quick work around, how can I move the lxc instances to the remaining nodes of the cluster? I copied the disk images, but I seem to be unable to copy the lxc.conf files, as they seem to be 'already present' (without being visible).