Cluster Nodes Can Connect to the Database, but Fail to Make a Cluster
This article applies to Confluence clustered 5.4 or earlier.
Symptoms
The typical cluster related fatal error occurs:
Fatal error in Confluence cluster: Database is being updated by an instance which is not part of the current cluster. You should check
network connections between cluster nodes, especially multicast traffic.
Diagnosis
The more typical causes for this error are described when occuring on a single node or when using Linux with IPv6. See Data Center Troubleshooting for more information.
In this case, even if the hosts are connected by crossover ethernet cable (so routing, router or switch issues are not possible) and the hosts can ping each other across the cable, the nodes are not able to make a cluster. Increase logging to DEBUG level - Confluence performs some UDP multicast testing. You will need to test that UDP traffic is allowed in the network.
Cause
UDP multicast is not being permitted across the network interface.
Resolution
Reconfigure the interface to permit UDP multicast. Alternatively, consider configuring a cluster for unicast traffic.