Adding a Node to Confluence Requires Prior Knowledge of IP

[Jeff Shepherd]

I am running Confluence Data Center 6.14.3 on AWS. Load balancer, two nodes, MySQL database backend.

Looking at documentation such as Adding and Removing Data Center Nodes it seems like all I need to do is duplicate a node (in AWS this would be a new instance from an AMI) and start it up.

However, the confluence.cluster.peers property in <confluence-home>/confluence.cfg.xml seems to be extremely important.

1. It seems to specify the nodes by IP address and it is a file found on each node.
2. If I omit the line, Confluence won't start up,
3. Not specifying the IP of each node in the cluster seems to cause problems as well

Does this mean if I add a node, I need to learn the IP address of the node and update confluence.cfg.xml on all my nodes? This does not lend itself to automation very well. Is there a way of adding a node without updating all the other nodes?

[Jeff Shepherd · Accepted solution]

Ah, my nodes are TCP/IP instead of Multicast

• https://confluence.atlassian.com/doc/change-node-discovery-from-multicast-to-tcp-ip-792297728.html