Bug in the master server on node reconnections

[gallagth]

Using the latest 0.2.1 release on OS X, MasterServer fails to correctly route nodes when they reconnect. To reproduce:

1. Compile and launch this master node:

import org.ros.RosCore;

import java.util.concurrent.TimeUnit;

/**
 * Created by thomas on 26/02/15.
 */
public class Main {

    public static void main(String [] args) {
        RosCore mRosCore = RosCore.newPublic(11311);
        mRosCore.start();
        try {
            mRosCore.awaitStart(5, TimeUnit.SECONDS);
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
       System.out.println("Ros core started");
       while (true);
    }
}

2. Launch the Talker and Listener nodes in the rosjava_tutorial_pubsub package, everything works as expected

3. Kill both nodes and relaunch them, they fail to connect to the master. The master throws:

SEVERE: Failed to invoke method registerPublisher in class org.ros.internal.node.xmlrpc.MasterXmlRpcEndpointImpl

And the nodes display:

SEVERE: Exception caught while communicating with master.
java.lang.RuntimeException: org.apache.xmlrpc.XmlRpcException: Failed to invoke method registerPublisher in class org.ros.internal.node.xmlrpc.MasterXmlRpcEndpointImpl: org/apache/commons/httpclient/HttpException

[toddsampson]

We are seeing the same thing. Any update on the status of the bug?