Assume there are four instances of WildFly 8 running on two machines within the same network. Let's name them A, B, C and D. A, B runs on machine1 and C, D on machine2.
In case of UDP clustering it's working fine with multicast address.
Using TCP I face some issues in clustering. I want A, C to be considered one and B, D to be considered one. Consider A, C as transaction servers and B, D as report servers. A, C have cache name
lrpcache and B, D have cache named
lrpcache1 and started with
ha profile. When I start the server A, C is looking for the cache in B, D as they are in same cluster.
When I connect all I get the following warning:
Discarding discovery request for cluster 'lrpcache1' from 54a568fd-b07a-dc68-01f3-0d978debf697; our cluster name is 'lrpcache'. Please separate your clusters cleanly.
I want A, C to work independent from B, D.
Servers were started as
standalone -c standalone-ha.xml -b 192.x.x.x.
The default TCP stack still uses multicast for discovery (via MPING). You should make sure the jgroups-mping socket binding for each group of servers uses a distinct multicast address/port.