[ http://jira.jboss.com/jira/browse/EJBTHREE-790?page=all ] Ortwin Glück updated EJBTHREE-790: ---------------------------------- Description: For ejb3-entity-cache-service.xml and ejb3-clustered-sfsbcache-service.xml the ClusterName should include the partition name. Here is why: We have 3 test instances and the two production cluster nodes within the same subnet (VLAN). The test instances have unique partition names (TEST1, TEST2, TEST3) and the production nodes share a common partition name (PROD). In the default configuration all five nodes would form a cluster for the entity and SFSB caches. This would severely disrupt data consistency! We have solved this by including the partition name in the cluster name in the cache MBeans: <attribute name="ClusterName">EJB3-entity-cache-${jboss.partition.name:DefaultPartition}</attribute> was: For ejb3-entity-cache-service.xml and ejb3-clustered-sfsbcache-service.xml the ClusterName should include the partition name. Here is why: We have 3 test instances and the two production cluster nodes within the same subnet (VLAN). The test instances have unique partition names (TEST1, TEST2, TEST3) and the production nodes share a common partition name (PROD). In the default configuration all five nodes would form a cluster for the entity and SFSB clusters. This would severely disrupt data consistency! We have solved this by including the partition name in the cluster name in the cache MBeans: <attribute name="ClusterName">EJB3-entity-cache-${jboss.partition.name:DefaultPartition}</attribute> -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/EJBTHREE-790?page=comments#action_12347222 ] Ortwin Glück commented on EJBTHREE-790: --------------------------------------- We actually have several instances on a (test) machine, using different ports for (not only) the caches. Your comment suggests that it's not enough that the caches have differen cluster names to isolate them. What's the catch? -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/EJBTHREE-790?page=all ] Brian Stansberry updated EJBTHREE-790: -------------------------------------- Fix Version/s: EJB 3.0 RC11 - FD -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/EJBTHREE-790?page=all ] Brian Stansberry closed EJBTHREE-790. ------------------------------------- Fix Version/s: AS 5.0.0.CR1 (was: HEAD) Resolution: Done The bit about including the partition name in the JGroups cluster name was fixed long ago. At this point I'm not going to do anything about the bit I mentioned on analyzing the bean's partition and identifying the correct cache from that. People need to specify the cache's properly (via @CacheConfig) if they deploy more than one partition. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira