Configuration distribuée de JMeter

When a single machine isn't enough to generate the required load, we need distributed testing. This means using multiple machines to simulate a massive number of users.

JMeter supports this by allowing you to run a test across several remote engines, all controlled by one master machine.

JMeter uses a Master-Slave architecture for distributed testing.

• Master: Your local machine, where you design the test plan and initiate the test. It collects results from the slaves.
• Slave (Engine): Remote machines that execute the test plan and generate the actual load against your target application.

Before setting up, ensure:

• Java: All master and slave machines have the same version of Java installed.
• JMeter: The same JMeter version is installed on all machines.
• Network: All machines are on the same network and can communicate with each other.
• Firewall: Ports are open for communication (default 1099, plus others if specified).

The master machine needs to know about its slaves. Open jmeter.properties (in JMeter's bin folder) on the master machine.

Find the remote_hosts property and list the IP addresses or hostnames of your slave machines, separated by commas.

On each slave machine, you typically don't need to modify remote_hosts. However, you might need to ensure RMI (Remote Method Invocation) communication is correctly set up.

If you have multiple network interfaces, set java.rmi.server.hostname to the slave's IP address in jmeter.properties. This tells the master which IP to connect to.

Once configured, start the JMeter server on each slave machine. Navigate to JMeter's bin directory and run the jmeter-server script.

This command starts the JMeter engine, making it ready to receive test plans from the master.

JMeter's distributed testing relies on RMI, which uses specific ports for communication. By default, RMI uses port 1099.

Ensure that port 1099 (or any custom RMI port you configure) is open in both directions (inbound/outbound) on the firewalls of all master and slave machines.

From the master machine, you can start the test using the command line. Use the -R flag followed by the slave IP addresses to specify which remote engines to use.

The test plan (JMX file) will be sent to the specified remote engines, which will then execute it concurrently.

You can also initiate a distributed test directly from the JMeter GUI on the master machine.

Go to Run > Remote Start and select the IP addresses of the slave machines you want to use. This offers a visual way to manage your remote engines.

Great job! You've learned how to configure JMeter for distributed load testing.

• We covered the Master-Slave architecture.
• You now know how to configure jmeter.properties on both master and slave machines.
• We discussed starting slave engines and running tests remotely from the master.
• Remember firewall considerations for smooth communication!

This setup allows you to simulate massive user loads, far beyond what a single machine can handle.