Examples of com.sleepycat.je.rep.impl.node.FeederManager

com.sleepycat.je.rep.impl.node.FeederManager

FeedManager is responsible for the creation and management of the Feeders used to respond to connections initiated by a Replica. runfeeders() is the central loop that listens for replica connections and manages the lifecycle of individual Feeders. It's re-entered each time the node becomes a Master and is exited when its status changes. There is a single instance of FeederManager that is created for a replication node. There are many instances of Feeders per FeederManager. Each Feeder instance represents an instance of a connection between the node serving as the feeder and the replica. Note that the FeederManager and the Replica currently reuse the Replication node's thread of control. When we implement r2r we will need to revisit the thread management to provide for concurrent operation of the FeederManger and the Replica.

                 * Re-informing when the node is a master is just an
                 * optimization, it does not impact correctness. Further
                 * minimize network traffic by trying just the nodes that are
                 * currently disconnected.
                 */
                final FeederManager feederManager = repNode.feederManager();
                final Set<String> active = feederManager.activeReplicas();
                active.add(repNode.getNodeName());


                final Set<InetSocketAddress> learners =
                    new HashSet<InetSocketAddress>();
                for (RepNodeImpl rn : repGroup.getAllMembers(null)) {

                 */
                return;
            }
            /* Snapshot the state to be used in the error message */
            final String dumpState = repImpl.dumpFeederState();
            final FeederManager feederManager =
                repImpl.getRepNode().feederManager();
            int currentFeederCount =
                feederManager.getNumCurrentAckFeeders(txn.getCommitVLSN());
            /*
             * Repeat the check to ensure that acks have not been received in
             * the time between the completion of the await() call above and
             * the creation of the exception message. This tends to happen when
             * there are lots of threads in the process thus potentially

Examples of com.sleepycat.je.rep.impl.node.FeederManager

Related Classes of com.sleepycat.je.rep.impl.node.FeederManager