检测到修复网关的连接丢失？（快速解决）-Detectinglossofconnectiontofixgateway?(QuickFix)

4  

The best way to solve this would probably be to decrease your Heartbeat Interval so you know sooner. I don't know of any messages that fire for a loss of TCP connection, but I don't think QuickFix is listening for OS events either. Although, it would likely flow through the fromAdmin event if there were such a message.

解决这个问题的最佳方法可能是降低心跳间隔,以便更快地了解。我不知道有任何消息会因为丢失TCP连接而触发,但我不认为QuickFix正在监听OS事件。虽然如果有这样的消息,它可能会流经fromAdmin事件。

Did you post your question to the QuickFix DL?

您是否将问题发布到QuickFix DL?

2  

It's possible that the fix engine you're using doesn't call back when TCP is disconnected, or it calls back on something other than onLogout. Since you're using fix, I would guess it forces a logout due to missed heartbeats.

您正在使用的修复引擎可能在TCP断开连接时不回调,或者回调onLogout以外的其他内容。由于您正在使用修复程序,我猜它会因为错过心跳而强制注销。

Quick way would be to look into the code and check where socket close is being handled, and what path executes when this happens.

快速的方法是查看代码并检查正在处理套接字关闭的位置,以及发生这种情况时执行的路径。

1  

TCP itself comes with a native heartbeat mechanism called SO_KEEPALIVE. Problem is that the default interval for this heartbeat can be as high as 2 hours. This is configured at the OS level. So in theory you can turn on SO_KEEPALIVE, configure a reasonable heartbeat interval at the OS level and be happy. However, because as said this is very OS dependent, most applications choose to implement heartbeating at the application level, and FIX is no exception. Decreasing your FIX heartbeat interval is the way to go here, especially if you are relying on cancel-on-disconnect and extra seconds of undetected connection loss can lead to unwanted order executions. A FIX gateway implemented on top of any fix engine should support heartbeating configuration out-of-the-box. Take a look on CoralGateway for an example. (Disclaimer: I am one of the developers of CoralGateway)

TCP本身带有一个名为SO_KEEPALIVE的本机心跳机制。问题是此心跳的默认间隔可能高达2小时。这是在OS级别配置的。因此理论上你可以打开SO_KEEPALIVE,在操作系统级别配置一个合理的心跳间隔并且很开心。但是,因为如上所述这是非常依赖操作系统的,大多数应用程序选择在应用程序级别实现心跳,并且FIX也不例外。减少FIX心跳间隔是这里的方法,特别是如果您依赖取消断开连接,并且未检测到连接丢失的额外秒数可能导致不必要的订单执行。在任何修复引擎之上实现的FIX网关应该支持开箱即用的心跳配置。以CoralGateway为例。 (免责声明:我是CoralGateway的开发者之一)