RabbitMQ---8、连接断开处理-断线重连

本文转载于：https://www.itsvse.com/thread-4636-1-1.html；

参考文献：http://www.likecs.com/show-29874.html；https://stackoverflow.com/questions/41279186/guaranteed-publishing-of-messages-on-rabbitmq-on-network-loss;

Rabbitmq 官方给的NET consumer示例代码如下，但使用过程，会遇到connection断开的问题，一旦断开，这个代码就会报错，就会导致消费者或者生产者挂掉。

下图是生产者发送消息，我手动停止了rabbitmq，然后又重新启动了rabbitmq，大概等启动成功以后，为了防止服务没有完全启动，我又等待了10秒钟

服务完全启动成功以后，我尝试重新发送一些消息，报错，如下：

************** 异常文本 **************
RabbitMQ.Client.Exceptions.AlreadyClosedException: Already closed: The AMQP operation was interrupted: AMQP close-reason, initiated by Peer, code=320, text="CONNECTION_FORCED - broker forced connection closure with reason 'shutdown'", classId=0, methodId=0, cause=
   在 RabbitMQ.Client.Impl.SessionBase.Transmit(Command cmd)
   在 RabbitMQ.Client.Impl.ModelBase.ModelSend(MethodBase method, ContentHeaderBase header, Byte[] body)
   在 RabbitMQ.Client.Impl.ModelBase.BasicPublish(String exchange, String routingKey, Boolean mandatory, IBasicProperties basicProperties, Byte[] body)
   在 RabbitMQ.Client.Impl.ModelBase.BasicPublish(String exchange, String routingKey, IBasicProperties basicProperties, Byte[] body)
   在 rabbitMQ_Publish.Form1.button1_Click(Object sender, EventArgs e) 位置 C:\project\my\RabbitMQ-demo\rabbitMQ-Publish\Form1.cs:行号 37
   在 System.Windows.Forms.Control.OnClick(EventArgs e)
   在 System.Windows.Forms.Button.OnClick(EventArgs e)
   在 System.Windows.Forms.Button.PerformClick()
   在 System.Windows.Forms.Form.ProcessDialogKey(Keys keyData)
   在 System.Windows.Forms.TextBoxBase.ProcessDialogKey(Keys keyData)
   在 System.Windows.Forms.Control.PreProcessMessage(Message& msg)
   在 System.Windows.Forms.Control.PreProcessControlMessageInternal(Control target, Message& msg)
   在 System.Windows.Forms.Application.ThreadContext.PreTranslateMessage(MSG& msg)

<ignore_js_op>

那么如何会异常恢复呢？或者说断线重连呢？

RabbitMQ NET Client的源码，研究发现一种自动的错误恢复机制 AutomaticRecoveryEnabled = true 使用方式如下

1. var factory = new ConnectionFactory() { HostName = "localhost", AutomaticRecoveryEnabled = true };

复制代码

具体的恢复机制如下

1.在AutoRecoveringConnection初始化时，在链接关闭事件委托上增加断开处理

1. public void init()
2. {
3. m_delegate = new Connection(m_factory, false, m_factory.CreateFrameHandler());
4. AutorecoveringConnection self = this;
5. EventHandler<ShutdownEventArgs> recoveryListener = (_, args) =>
6. {
7. lock (recoveryLockTarget)
8. {
9. if (ShouldTriggerConnectionRecovery(args))
10. {
11. try
12. {
13. self.BeginAutomaticRecovery();
14. }
15. catch (Exception e)
16. {
17. // TODO: logging
18. Console.WriteLine("BeginAutomaticRecovery() failed: {0}", e);
19. }
20. }
21. }
22. };
23. lock (m_eventLock)
24. {
25. ConnectionShutdown += recoveryListener;
26. if (!m_recordedShutdownEventHandlers.Contains(recoveryListener))
27. {
28. m_recordedShutdownEventHandlers.Add(recoveryListener);
29. }
30. }
31. }

复制代码

观察调用的方式BeginAutomaticRecovery，可以看到这个方法内部调用了PerformAutomaticRecovery方法。我们直接看这个方法的内容，其中第一个调用的是方法RecoverConnectionDelegate

1. protected void PerformAutomaticRecovery()
2. {
3. lock (recoveryLockTarget)
4. {
5. RecoverConnectionDelegate();
6. RecoverConnectionShutdownHandlers();
7. RecoverConnectionBlockedHandlers();
8. RecoverConnectionUnblockedHandlers();
9. RecoverModels();
10. if (m_factory.TopologyRecoveryEnabled)
11. {
12. RecoverEntities();
13. RecoverConsumers();
14. }
15. RunRecoveryEventHandlers();
16. }
17. }

复制代码

这个方法中调用的是

1. protected void RecoverConnectionDelegate()
2. {
3. bool recovering = true;
4. while (recovering)
5. {
6. try
7. {
8. m_delegate = new Connection(m_factory, false, m_factory.CreateFrameHandler());
9. recovering = false;
10. }
11. catch (Exception)
12. {
13. // TODO: exponential back-off
14. Thread.Sleep(m_factory.NetworkRecoveryInterval);
15. // TODO: provide a way to handle these exceptions
16. }
17. }
18. }

复制代码

可以看出，它是执行了死循环，直到连接重新打开，当然，如果遇到异常，它会调用Thread.Sleep来等待一下，然后再次执行连接恢复。