1 Prepare the machine
N+1 machines, referred to as the client
N as slave, 1 as controller
Note: The slave initiates the machine as a virtual user
2 configuration
1 Turn off all slave firewalls --- slave
2 All clients should be in the same subnet. --- Client
3 Make sure jMeter can access this server ---Ping --client can access the server
4 Make sure that the jMeter versions of each client are consistent, different versions of JMeter may not work together. --- The Jmeter version of all clients should be the same
5 Set the slave IP ( if the controller also wants to generate a virtual user, add: localhost:1099 and need to start jmeter-server.bat )
If the environment variable settings have been set, then in the control system, open the jmeter.properties file in the bin directory , and in the " remote_hosts " line, add the slave's ip, such as reomte_hosts=192.168.0.10,192: 1099,168,0.13:1099, then start jmeter , import the test plan you want to execute in the control system, pay attention to adding the port number here, otherwise the startup will fail ( the proof of version 2.6 , it is not necessary ) .
6 Start the slave " jmeter-server.bat "
If you want to confirm whether the slave system has been started, you can use Notepad to open the jmeter file in the bin directory . If it starts normally, you should see the following log. Sometimes it will report that the "ApacheJMeter_core.jar" package cannot be found when starting from the machine. Solution: Configure the jmeter environment variable in the user variable column: "JMETER_HOME" value is an absolute path"D:\software\test\jmeter-2-6 -en-win\apache-jmeter-2.6"
7 Check the number of tcp connections. The purpose is to roughly look at the virtual users generated by the client. Use nopad+ to search for ip .
Netstat –n > e:/log.txt