we work in jboss 4.3 cluster environment for an application that manage contents.
We use lucene 4 manage contents
We have a problem in call replication from node1 of cluster to node2 of cluster (terracotta 2.7.3)

On the node1 the call of terracotta (dmi_stage) are between the local calls

Code:

2010-12-20 11:03:04,355  WorkerThread#1[10.102.40.94:33002] READLOCK START 
 2010-12-20 11:03:04,355  WorkerThread#1[10.102.40.94:33002] READLOCK END 0 
 2010-12-20 11:03:04,355  WorkerThread#1[10.102.40.94:33002] RELEASEREADLOCK START 
 2010-12-20 11:03:04,355  WorkerThread#1[10.102.40.94:33002] RELEASEREADLOCK END 0 
 2010-12-20 11:03:04,355  WorkerThread(dmi_stage,0) READLOCK START 
 2010-12-20 11:03:04,355  WorkerThread(dmi_stage,0) READLOCK END 0 
 2010-12-20 11:03:04,355  WorkerThread(dmi_stage,0) RELEASEREADLOCK START 
 2010-12-20 11:03:04,355  WorkerThread(dmi_stage,0) RELEASEREADLOCK END 0 
 2010-12-20 11:03:04,437  WorkerThread#4[10.102.46.185:33489] READLOCK START 
 2010-12-20 11:03:04,437  WorkerThread#4[10.102.46.185:33489] READLOCK END 0 
 2010-12-20 11:03:04,437  WorkerThread#4[10.102.46.185:33489] RELEASEREADLOCK START 
 2010-12-20 11:03:04,437  WorkerThread#4[10.102.46.185:33489] RELEASEREADLOCK END 0 
 

On the second node of cluster, on the other way, the calls of terracotta are always replicated at the end

Code:

2010-12-20 11:03:05,395 WorkerThread#0[10.102.46.186:32999] RELEASEWRITELOCK START 
 2010-12-20 11:03:05,396 WorkerThread#0[10.102.46.186:32999] RELEASEWRITELOCK END 1 
 2010-12-20 11:03:05,396 WorkerThread#0[10.102.46.186:32999] RELEASEREADLOCK START 
 2010-12-20 11:03:05,396 WorkerThread#0[10.102.46.186:32999] RELEASEREADLOCK END 0 
 2010-12-20 11:03:07,512 WorkerThread(dmi_stage,0) READLOCK START 
 2010-12-20 11:03:07,512 WorkerThread(dmi_stage,0) READLOCK END 0 
 2010-12-20 11:03:07,512 WorkerThread(dmi_stage,0) RELEASEREADLOCK START 
 2010-12-20 11:03:07,512 WorkerThread(dmi_stage,0) RELEASEREADLOCK END 0 
 2010-12-20 11:03:07,512 WorkerThread(dmi_stage,0) WRITELOCK START 
 2010-12-20 11:03:07,512 WorkerThread(dmi_stage,0) WRITELOCK END 0 

The problem is that the calls on node1 of cluster wait that calls will end on the node2 so the client wait the end of replication

Could you help me ?

Alessandro