I'm currently using infinispan with following configuration:

-    transactional mode (optimistic locking - repeatable read)

-    only one node of grid assigned to perform put/remove operations (hereafter called master node), and all remaining grid nodes able to perform only read operations (hereafter called read nodes). I have a low total number of grid nodes.

-    synchronous replycated mode

Now, I would like to know how I have to change my infinispan configuration in order to work with all following characteristics at the same time:

-    maintain transactional mode (optimistic locking - repeatable read)

-    maintain only one master node and all other as read nodes

-    asynchronous replycated mode

-   in case of an error of data replication occurred only in one read node X (e.g. ISPN000220: Problems un-marshalling remote command from byte buffer: java.io.InvalidObjectException: Attempt to read a backreference with an invalid ID), it is ensured that data are properly replicated in all other nodes, and then it will be infinispan to retry data replication on read node X (once it has been re-deployed with fixed error)

-   finally, I need to intercept in the master node all Exceptions related to possibly errors occurred during execution of asynchronous transactions, in order to store error messages in a dedicated db table. How can I intercept these Exceptions?

Many thanks

Francesco Sclano