转载

jmeter压力测试(踩坑)报错:java.net.BindException: Address already in use: connect

最近在写搜索接口服务，写完接口进行压力测试，但是在长期线程较高的进行测试时会报如下错误：

java.net.BindException: Address already in use: connect
        at java.net.DualStackPlainSocketImpl.connect0(Native Method)
        at java.net.DualStackPlainSocketImpl.socketConnect(Unknown Source)
        at java.net.AbstractPlainSocketImpl.doConnect(Unknown Source)
        at java.net.AbstractPlainSocketImpl.connectToAddress(Unknown Source)
        at java.net.AbstractPlainSocketImpl.connect(Unknown Source)
        at java.net.PlainSocketImpl.connect(Unknown Source)
        at java.net.SocksSocketImpl.connect(Unknown Source)
        at java.net.Socket.connect(Unknown Source)
        at org.apache.http.conn.socket.PlainConnectionSocketFactory.connectSocket(PlainConnectionSocketFactory.java:75)
        at org.apache.http.impl.conn.DefaultHttpClientConnectionOperator.connect(DefaultHttpClientConnectionOperator.java:142)
        at org.apache.jmeter.protocol.http.sampler.HTTPHC4Impl$JMeterDefaultHttpClientConnectionOperator.connect(HTTPHC4Impl.java:331)
        at org.apache.http.impl.conn.PoolingHttpClientConnectionManager.connect(PoolingHttpClientConnectionManager.java:373)
        at org.apache.http.impl.execchain.MainClientExec.establishRoute(MainClientExec.java:394)
        at org.apache.http.impl.execchain.MainClientExec.execute(MainClientExec.java:237)
        at org.apache.http.impl.execchain.ProtocolExec.execute(ProtocolExec.java:185)
        at org.apache.http.impl.execchain.RetryExec.execute(RetryExec.java:89)
        at org.apache.http.impl.execchain.RedirectExec.execute(RedirectExec.java:110)
        at org.apache.http.impl.client.InternalHttpClient.doExecute(InternalHttpClient.java:185)
        at org.apache.http.impl.client.CloseableHttpClient.execute(CloseableHttpClient.java:83)
        at org.apache.jmeter.protocol.http.sampler.HTTPHC4Impl.executeRequest(HTTPHC4Impl.java:832)
        at org.apache.jmeter.protocol.http.sampler.HTTPHC4Impl.sample(HTTPHC4Impl.java:570)
        at org.apache.jmeter.protocol.http.sampler.HTTPSamplerProxy.sample(HTTPSamplerProxy.java:67)
        at org.apache.jmeter.protocol.http.sampler.HTTPSamplerBase.sample(HTTPSamplerBase.java:1231)
        at org.apache.jmeter.protocol.http.sampler.HTTPSamplerBase.sample(HTTPSamplerBase.java:1220)
        at org.apache.jmeter.threads.JMeterThread.doSampling(JMeterThread.java:622)
        at org.apache.jmeter.threads.JMeterThread.executeSamplePackage(JMeterThread.java:546)
        at org.apache.jmeter.threads.JMeterThread.processSampler(JMeterThread.java:486)
        at org.apache.jmeter.threads.JMeterThread.run(JMeterThread.java:253)
        at java.lang.Thread.run(Unknown Source)

排除问题：

首先先查看服务器的日志，发现没有报错。然后查看nginx数据，发现请求数和测试发出的请求数不一致，服务器接收到的少，就想到丢失请求。后来经过查找资料了解是windows 机器的问题，原因：windows提供给TCP/IP链接的端口为 1024-5000，并且要四分钟来循环回收它们，就导致我们在短时间内跑大量的请求时将端口占满了，导致如上报错。

解决办法（在jmeter所在服务器操作）：

1.cmd中输入regedit命令打开注册表；
2.在 HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters右键Parameters；
3.添加一个新的DWORD，名字为MaxUserPort；
4.然后双击MaxUserPort，输入数值数据为65534，基数选择十进制；
5.完成以上操作，务必重启机器，问题解决。
解决后的测试结果就不再报错：