properly test durable binding recovery

we create the durable queue on the secondary node (if available),
which gets fully restarted, rather than the primary node, on which
only the rabbit app is restarted.

Author: Matthias Radestock

test/src/com/rabbitmq/client/test/functional/BindingLifecycle.java

@@ -31,6 +31,8 @@
 
 package com.rabbitmq.client.test.functional;
 
+import com.rabbitmq.client.Channel;
+import com.rabbitmq.client.Connection;
 import com.rabbitmq.client.GetResponse;
 import com.rabbitmq.client.QueueingConsumer;
 
@@ -53,6 +55,29 @@ public class BindingLifecycle extends PersisterRestartBase {
     protected static final String X = "X-" + System.currentTimeMillis();
     protected static final String K = "K-" + System.currentTimeMillis();
 
+    /**
+     * Create a durable queue on secondary node, if possible, falling
+     * back on the primary node if necessary.
+     */
+    @Override protected void declareDurableQueue(String q)
+        throws IOException
+    {
+        Connection connection;
+        try {
+            connection = connectionFactory.newConnection("localhost", 5673);
+        } catch (IOException e) {
+            super.declareDurableQueue(q);
+            return;
+        }
+
+        Channel channel = connection.createChannel();
+
+        channel.queueDeclare(q, true);
+
+        channel.abort();
+        connection.abort();
+    }
+
     /**
      *   Tests whether durable bindings are correctly recovered.
      */