1 Reply Latest reply on Jan 24, 2007 4:49 PM by Luc DEW

    Behaviour of the cache in case of failure of a node

    Luc DEW Newbie


      I have a newbie question. I'd like to know how the Pojo cache (1.4.1) behaves in case of failure of a node in a 2 nodes cluster.

      Our Pojo cache is configured to perform synchronous replication and to be transactional (the JTA transaction manager of our application server is used).

      When a node is down and we call transactional methods that update objects in the cache, everything works fine without any error.

      So i guess there's a node state detection implemented as a multicast heartbeat or something equivalent. But:
      - what would be the behavior if an update of a cached object is done on a node and at the same time the other node crashes or can not be accessed ? Reading the docs it says that an exception is thrown (of which type ?)
      and the transaction is rolled back.
      - How Jboss cache makes the difference between a shutdown node and node that can not be reached (synchronisation can not be performed) due to network failure or a Jvm freeze ?

      Thanks in advance.