When setting up LiveCycle clusters, if the cluster nodes are running on hosts that have multiple NIC’s (multi-homed), then the following additional JVM argument must be configured along with the configurations provided in the Cluster documentation
-Dgemfire.bind-address
For example: If there are 2 nodes in a cluster and they are running on hosts with 2 NIC’s -
• Node 1 : NIC1=10.217.140.41; NIC2=192.168.131.48 – where LiveCycle is running on NIC1
• Node 2 : NIC1=10.217.140.43; NIC2=192.168.131.49 – where LiveCycle is running on NIC1
Then for:
• Node 1 the additional JVM argument to be configured will be
-Dgemfire.bind-address=10.217.140.41
• Node 2 the additional JVM argument to be configured will be
-Dgemfire.bind-address=10.217.140.43
If this parameter is not configured then in the Gemfire logs you may see that the gemfire.jar from the LiveCycle Core EAR may try to use the IP of the other NIC on the host. For the above example when using TCP LOcators as the cluster caching mechanism you may end up seeing messages as below -
————————————
info 2011/08/04 09:46:31.003 CEST GemfireCacheAdapter <main> tid=0×8] GemFire P2P Listener started on tcp:///192.168.131.248:58332 <– This shoudl be tcp:///10.217.140.41 instead for correct functioning of the cluster.
————————————
Configuring the above JVM parameter will resolve this issue.
