3 Replies Latest reply on Sep 15, 2010 9:53 PM by Changgeng Li

    "Locking a single remote node" in 4.2.0ALPHA1

    Changgeng Li Newbie

      I think before accquiring a remote lock, the current implementation will still try to acquire a local lock, is that correct?


      Could we have an option to entirely disable the local lock? This will be very helpful to avoid deadlock(local vs. remote) when there's a heavy concurrent write load on a specific key.



      Also I create a testcase with the attached cluster.xml, and it seems there's some bug that I got the following exception.


      Caused by: java.lang.RuntimeException: Failure to marshal argument(s)
              at org.infinispan.remoting.transport.jgroups.CommandAwareRpcDispatcher$ReplicationTask.marshallCall(CommandAwareRpcDispatcher.java:242)
              at org.infinispan.remoting.transport.jgroups.CommandAwareRpcDispatcher$ReplicationTask.call(CommandAwareRpcDispatcher.java:266)
              at org.infinispan.remoting.transport.jgroups.CommandAwareRpcDispatcher.invokeRemoteCommands(CommandAwareRpcDispatcher.java:120)
              ... 42 more
      Caused by: org.infinispan.marshall.NotSerializableException: org.infinispan.transaction.xa.DistDldGlobalTransaction
      Caused by: an exception which occurred:
              in object org.infinispan.transaction.xa.DistDldGlobalTransaction@19a1027a
              in object org.infinispan.commands.tx.RollbackCommand@19a1027a