Dynamic Queue Coordinator not failing over

  • Perhaps somebody has some insight into the following problem.

    We have producers located on multiple nodes A,B that send their results to node C using queue Q. When we send the first message from node A, MR tells us that it is starting the queue coordinator on node A, and everything works OK. However if node A crashes, the messages sent from node B are stuck at the send() stage until I either restart the application on node B, or restart node A. Does anyone have any ideas?

       Thanks in advance