Dynamically changing wsrep_cluster_name causes non-Primary mode

Changing the `wsrep_cluster_name` causes the cluster node(s) to go into non-Primary mode. I tested this on a 2-node cluster where both nodes went into non-Primary mode causing an outage. This might not happen when changing this parameter on a 1 node of a 3-node cluster.

According to Percona fixing the same issue in 2016 (https://bugs.launchpad.net/percona-xtradb-cluster/+bug/1620439) changing the wsrep_cluster_name reloads the provider library. Therefore Percona changed the variable to not be dynamic anymore.

Please at least change the documentation to mention that it is unsafe to change the configuration parameter.

Next to that it could be worth to mention how the reloaded provider can be made Primary again. I restarted the nodes thus did not confirm the procedure, which should be something like this:

1. Shutdown all nodes but 1
2. Change the configuration file on all nodes
3. Dynamically change the configuration parameter
4. Make that node primary again
5. Start the other nodes again

Thirdly, you may want to make the parameter dynamic only when there is 1 node in the cluster, and make sure that this 1 node does not get non-Primary from changing it.

The reason you want to be able to change the cluster name is to avoid other nodes that have the default cluster name as well to be able to connect to the cluster.

Thank you!