JMS Colocated Failover Recover Only Example

This example demonstrates how you can colocate live and backup servers in the same VM. We do this by creating an HA Policy that is colocated. colocated means that backup servers can be created and maintained by live servers on behalf of other requesting live servers. In this example we create a colocated shared store server that will scale down. That is it will not become live but scale down the journal to the colocated live server.

This example starts 2 live servers each with a backup server that backs up the other live server.

The first live server will be killed and the backup in the second will recover the journal and recreate its state in the live server it shares its VM with.

The following shows how to configure the backup, the backup strategy is set to SCALE_DOWN which means that the backup server will not fully start on fail over, instead it will just recover the journal and write it to its parent live server. Also notice we have over ridden some of the configuration since we want it to use the same journal as server1 since it is using shared store.

     <ha-policy template="COLOCATED_SHARED_STORE"/>

note that for this HA policy we use a template that will use some sensibe settings, in this case this includes setting scale down to true. Also note that since we dont specify a scale down connector it will use most appropriate from the list of available connectors which in this case is the first INVM connector

One other thing to notice is that the cluster connection has its reconnect attempts set to 5, this is so it will disconnect instead of trying to reconnect to a backup that doesn't exist.

Example step-by-step

To run the example, simply type mvn verify from this directory

  1. Get an initial context for looking up JNDI for both servers
    2.            
    initialContext1 = getContext(1);
    initialContext = getContext(0);
  2. Look up the JMS resources from JNDI
    3.            
    Queue queue = (Queue)initialContext.lookup("/queue/exampleQueue");
    ConnectionFactory connectionFactory = (ConnectionFactory)initialContext.lookup("/ConnectionFactory");
    ConnectionFactory connectionFactory1 = (ConnectionFactory)initialContext1.lookup("/ConnectionFactory");
  3. Create a JMS Connections
    4.            
    connection = connectionFactory.createConnection();
    connection1 = connectionFactory1.createConnection();
  4. Create a *non-transacted* JMS Session with client acknowledgement
    5.            
    Session session = connection.createSession(false, Session.CLIENT_ACKNOWLEDGE);
    Session session1 = connection1.createSession(false, Session.CLIENT_ACKNOWLEDGE);
  5. Create a JMS MessageProducer
    6.            
    MessageProducer producer = session.createProducer(queue);
    MessageProducer producer1 = session1.createProducer(queue);
  6. Send some messages to both servers
    7.           
    for (int i = 0; i < numMessages; i++)
    {
      TextMessage message = session.createTextMessage("This is text message " + i);
      producer.send(message);
      System.out.println("Sent message: " + message.getText());
      message = session1.createTextMessage("This is another text message " + i);
      producer1.send(message);
      System.out.println("Sent message: " + message.getText());
    }
  7. Crash server #0, the live server
    8.           
    killServer(0);
  8. start the connection ready to receive messages
    9.            
    connection1.start();
  9. create a consumer
    10.            
    MessageConsumer consumer = session1.createConsumer(queue);
  10. Receive and acknowledge all of the sent messages, notice that they will be out of order, this is because they were initially round robined to both nodes then when the server failed were reloaded into the live server.
    11.            
    TextMessage message0 = null;
    for (int i = 0; i < numMessages * 2; i++)
    {
       message0 = (TextMessage)consumer.receive(5000);
       System.out.println("Got message: " + message0.getText());
    }
    message0.acknowledge();
  11. And finally (no pun intended), always remember to close your JMS resources after use, in a finally block. Closing a JMS connection will automatically close all of its sessions, consumers, producer and browser objects
    12.            
    finally
    {
       if (connection != null)
       {
       connection.close();
       }

       if (initialContext != null)
       {
       initialContext.close();
       }
       if (connection1 != null)
       {
       connection1.close();
       }

       if (initialContext1 != null)
       {
       initialContext1.close();
       }
    }