JBoss配置解决高并发连接异常问题(转)

这两天一个项目在做压力测试的时候，发现只要并发数超过250个，连续测试两轮就会有连接异常出现，测试轮数越多出现越频繁，异常日志如下：

Caused by: com.caucho.hessian.client.HessianConnectionException: 500: java.io.IOException: Error writing to server
    at com.caucho.hessian.client.HessianURLConnection.sendRequest(HessianURLConnection.java:142)
    at com.caucho.hessian.client.HessianProxy.sendRequest(HessianProxy.java:283)
    at com.caucho.hessian.client.HessianProxy.invoke(HessianProxy.java:170)
    at $Proxy168.sendOpenAcctInfo(Unknown Source)
    at sun.reflect.GeneratedMethodAccessor750.invoke(Unknown Source)
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25)
    at java.lang.reflect.Method.invoke(Method.java:597)
    at org.springframework.remoting.caucho.HessianClientInterceptor.invoke(HessianClientInterceptor.java:219)
    at org.springframework.aop.framework.ReflectiveMethodInvocation.proceed(ReflectiveMethodInvocation.java:171)
    at org.springframework.aop.framework.JdkDynamicAopProxy.invoke(JdkDynamicAopProxy.java:204)
    at $Proxy169.sendOpenAcctInfo(Unknown Source)
    at com.shine.web.bean.OpenAcctBeanImpl.sendOpenAcctInfo(OpenAcctBeanImpl.java:62)
    ... 32 more
Caused by: java.io.IOException: Error writing to server
    at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
    at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:39)
    at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:27)
    at java.lang.reflect.Constructor.newInstance(Constructor.java:513)
    at sun.net.www.protocol.http.HttpURLConnection$6.run(HttpURLConnection.java:1345)
    at java.security.AccessController.doPrivileged(Native Method)
    at sun.net.www.protocol.http.HttpURLConnection.getChainedException(HttpURLConnection.java:1339)
    at sun.net.www.protocol.http.HttpURLConnection.getInputStream(HttpURLConnection.java:993)
    at com.caucho.hessian.client.HessianURLConnection.sendRequest(HessianURLConnection.java:122)
    ... 43 more
Caused by: java.io.IOException: Error writing to server
    at sun.net.www.protocol.http.HttpURLConnection.writeRequests(HttpURLConnection.java:453)
    at sun.net.www.protocol.http.HttpURLConnection.writeRequests(HttpURLConnection.java:465)
    at sun.net.www.protocol.http.HttpURLConnection.getInputStream(HttpURLConnection.java:1047)
    at java.net.HttpURLConnection.getResponseCode(HttpURLConnection.java:373)
    at com.caucho.hessian.client.HessianURLConnection.sendRequest(HessianURLConnection.java:109)
    ... 43 more

一开始使用Error writing to server去网络上查找原因，发现基本都无关。

后来搜索hessian和spring兼容问题，发现spring2.5.6和hessian4.0.7不兼容。将hessian版本号降低到3.1.3，情况有好转，但测试10轮之后，异常又出现了。

通过内存监控，先排除虚拟机内存问题。虚拟机内存配置为-Xms1024m -Xmx1024m，监控下来发现实际占用内存不到一半。

然后通过netstat -na监控操作系统端口占用情况，发现端口占用高峰不到500个，这个原因也排除了。（测试服务器已经修改注册表将TIME_WAIT时间降低到30秒，所以基本不会出现端口占用问题）。

通过CPU监控，确认并发高峰时，CPU占用也不到50%。

种种迹象表明，这些都不是造成连接断开的原因，那到底瓶颈出现在哪里呢？

于是我们将目光转向JBoss的配置。

首先确认数据库的连接池配置，最大连接数设置为50，从前几轮都可以正常运行来看，数据库连接应该够用；

然后确认JBoss的线程池配置，发现默认配置如下：

<mbean code="org.jboss.util.threadpool.BasicThreadPool"
   name="jboss.system:service=ThreadPool">
   <attribute name="Name">JBoss System Threads</attribute>
   <attribute name="ThreadGroupName">System Threads</attribute>
   <!-- How long a thread will live without any tasks in MS -->
   <attribute name="KeepAliveTime">60000</attribute>
   <!-- The max number of threads in the pool -->
   <attribute name="MaximumPoolSize">10</attribute>
   <!-- The max number of tasks before the queue is full -->
   <attribute name="MaximumQueueSize">1000</attribute>
   <!-- The behavior of the pool when a task is added and the queue is full.
   abort - a RuntimeException is thrown
   run - the calling thread executes the task
   wait - the calling thread blocks until the queue has room
   discard - the task is silently discarded without being run
   discardOldest - check to see if a task is about to complete and enque
      the new task if possible, else run the task in the calling thread
   -->
   <attribute name="BlockingMode">run</attribute>
</mbean>

搜索了一下相关配置的说明，在进行高并发的时候，建议修改MaximumPoolSize的大小为并发数的125%。

由于我们测试的不是持续的并发，因此将线程池大小修改成200先测试了一下，发现并发数在300的时候可以正常运行，又将并发数修改到500，持续测试了6个小时，均没有发现异常。

现在比较好奇的是，为什么250个并发的时候就能一直不出错，超过250个并发，就会频繁出错，这个值和MaximumPoolSize的参数到底有什么联系呢？

http://blog.csdn.net/nicholas_lin/article/details/20639481

 http://wenku.baidu.com/link?url=eUQiTt73bQN_XBHVNpAhDnSMYfLdfqQXK1AF5Pp2dhTgBrO4nHaws7rEm8WZY5WVIiOEUaX5UQuuQTNCM9DrsNMjetboto1NnikLSEtzH6S