Clustering over multiple network interfaces

I have a happily running 3 node cluster, which is running firewalled to the hilt, no problems at all. I have now networked these 3 nodes to a 4th, but am using a VPN to do so, so each node has gained an extra IP address.

I want to use the VPN connections to create an emergency(!) 4th node, but when just reconfiguring this new node to access the other 3 with the new network addresses by updating wsrep_cluster_address for this node, I just get a timeout: it downloads galera.cache and grastate.dat, then i get the following in the error log:

2014-07-17 14:19:29 19672 [Note] WSREP: view((empty))
2014-07-17 14:19:29 19672 [ERROR] WSREP: failed to open gcomm backend connection: 110: failed to reach primary view: 110 (Connection timed out)
at gcomm/src/pc.cpp:connect():141
2014-07-17 14:19:29 19672 [ERROR] WSREP: gcs/src/gcs_core.c:gcs_core_open():202: Failed to open backend connection: -110 (Connection timed out)
2014-07-17 14:19:29 19672 [ERROR] WSREP: gcs/src/gcs.c:gcs_open():1291: Failed to open channel ‘MYCHANNEL’ at ‘gcomm://192.168.93.210,192.168.93.220,192.168.93.230,192.168.93.1’: -110 (Connection timed out)
2014-07-17 14:19:29 19672 [ERROR] WSREP: gcs connect failed: Connection timed out
2014-07-17 14:19:29 19672 [ERROR] WSREP: wsrep::connect() failed: 7
2014-07-17 14:19:29 19672 [ERROR] Aborting

2014-07-17 14:19:29 19672 [Note] WSREP: Service disconnected.
2014-07-17 14:19:30 19672 [Note] WSREP: Some threads may fail to exit.
2014-07-17 14:19:30 19672 [Note] Binlog end
2014-07-17 14:19:30 19672 [Note] /usr/sbin/mysqld: Shutdown complete

The new node is the 192.168.93.1, and .210, .220 and .230 are the new interfaces on the existing nodes. wsrep_cluster_address has been updated on 2 of the 3 existing nodes with the 192.168.93.1 address in addition to the 3 existing external IP addresses and restarted - .210 hasn’t.

Is what I’m trying to do possible? Do I need to add all 7 of the IP addresses, and if so, can it handle multiple connections to the same node ok?

Any suggestions gratefully received… am square eyed from docs!

Cheers,

Steve

Gah… it was the Galeria 2/3 thing that bit me! All running fine now, with the above configuration.