I have been configuring this oracle coherence more then 2 weeks. and I must say its been a pleasure. I love clustered things. they mostly does not work and hard to figure out. but oracle coherence is not that complicated.
one thing. in our production environment I could not use multi casters and convert configuration to unicast stuff. its a p2p based tcmp network and solid work. I can say oracle coherence is something like memcached. and its easy to figure out. because it can run under tomcat and deployed to weblogic servers with our application.
while production tests I saw that if you change any simple detail in tangosol-coherence-override.xml I need to restart weblogic instances. because when coherence started it could not change its configuration on the fly. I think if we used coherence jar out of our war it would be more flexible. but our cache system does not need to configured every one day or so. so we keep those xmls(tangosol-coherence-override.xml,coherence-cache-config.xml) inside our war.
another thing. if I add well-known-addresses to tangosol xml it trys to connect while starting the war. and if there is no localhost coherence does not look for it. so I need to change tangosol like this.
10.210.61.7
13220
10.210.59.231
13220
localhost
13220 30000 35000 /var/log/comusr/WLS_LOGS/coherence/coherence.log 9 com.oracle.coherence.environment.extensible.ExtensibleEnvironment java.lang.String coherence-cache-config.xml EE prod this configuration looks for localhost and runs well in redhat enterprise linux. coherence version:3.7.1.0