Using gRPC with JMeter: Writing Java Sampler Scripts for RPC Performance Testing

The article introduces the growing use of RPC frameworks such as Dubbo, Motan, Thrift, and especially gRPC, and explains how to perform load testing on RPC services using JMeter, which only supports Java‑based samplers.

Step 1: Prepare the RPC script – Create a Maven project, add gRPC and JMeter dependencies in pom.xml , place the .proto files (e.g., gnid.proto ) under a new source directory, run mvn install to generate the client and server code, and copy the generated code into the appropriate package.

Step 2: Write the Java sampler for JMeter – Implement a class that either extends AbstractJavaSamplerClient or implements JavaSamplerClient . Override the required methods as shown below:

public Arguments getDefaultParameters(): //设置可用参数的默认值；
public void setupTest(JavaSamplerContext arg0): //每个线程测试前执行一次，做一些初始化工作；
public SampleResult runTest(JavaSamplerContext arg0): //开始测试，从arg0参数可以获得参数值；
public void teardownTest(JavaSamplerContext arg0): //测试结束时调用；

In the sampler, initialize the long‑lived RPC connection in setupTest and close it in teardownTest to mimic production usage. Parameters can be added to distinguish multiple RPC interfaces.

Step 3: Package the script – Build the compiled sampler into a JAR and place it in JMeter’s \lib\ext\ directory.

Step 4: Run the test in JMeter – Create a Thread Group, add a Java Request sampler, select the custom sampler class, and configure the parameters defined in the code. Add assertions, listeners, and any needed parameterization, then start the test to generate load similar to HTTP testing.

Conclusion – The guide covers two key points: how to write RPC performance test scripts and how to execute them with JMeter’s Java sampler. Readers are encouraged to ask questions in the comments.