如何在JMeter中扩展聚合报告、摘要报告监听器以获取运行后数据？

Hey Krishna, great question! Extending JMeter's built-in reports to combine their data is totally doable, and it boils down to leveraging JMeter's existing listener architecture. Here's a step-by-step guide to make it happen:

Prerequisites #

First, make sure you have these set up:

• JDK 8+ (JMeter 5.x and above require this)
• An IDE (IntelliJ, Eclipse, etc.) for writing Java code
• JMeter's core dependencies (either via Maven/Gradle or by including JMeter's JARs from your installation)

Step 1: Set Up Your Project #

If using Maven, add these dependencies to your pom.xml to pull in JMeter's core libraries:

<dependencies>
    <dependency>
        <groupId>org.apache.jmeter</groupId>
        <artifactId>ApacheJMeter_core</artifactId>
        <version>5.6.3</version> <!-- Use the latest JMeter version -->
        <scope>provided</scope>
    </dependency>
    <dependency>
        <groupId>org.apache.jmeter</groupId>
        <artifactId>ApacheJMeter_reporters</artifactId>
        <version>5.6.3</version>
        <scope>provided</scope>
    </dependency>
</dependencies>

Step 2: Create Your Custom Listener Class #

JMeter's AggregateReport and SummaryReport both extend AbstractReport (from org.apache.jmeter.reporters), which handles the heavy lifting of collecting sample data. Your custom listener can extend this class too, then reuse the existing stats collection while adding support for outputting all metrics from both reports.

Here's a basic skeleton:

package com.yourorg.jmeter.listeners;

import org.apache.jmeter.reporters.AbstractReport;
import org.apache.jmeter.samplers.SampleEvent;

import java.io.PrintWriter;

public class CombinedAggregateSummaryReport extends AbstractReport {

    // The AbstractReport already has a protected "stats" variable that collects all necessary data
    // We just need to adjust how we output this data

    @Override
    public void sampleOccurred(SampleEvent event) {
        // Let the parent class handle data collection
        super.sampleOccurred(event);
        // Add custom sample processing here if needed
    }

    @Override
    public void printResults(PrintWriter out) {
        // First print Summary Report-style high-level metrics
        printSummaryMetrics(out);
        // Then print Aggregate Report-style detailed percentile metrics
        printAggregateMetrics(out);
    }

    private void printSummaryMetrics(PrintWriter out) {
        out.println("=== Summary Report Metrics ===");
        out.printf("Total Samples: %d%n", stats.getTotal());
        out.printf("Error Count: %d (%.2f%%)%n", stats.getErrorCount(), stats.getErrorPercentage());
        out.printf("Average Response Time: %.2f ms%n", stats.getMean());
        out.printf("Throughput: %.2f requests/sec%n", stats.getRate());
        out.println("------------------------------");
    }

    private void printAggregateMetrics(PrintWriter out) {
        out.println("=== Aggregate Report Metrics ===");
        out.printf("Median: %.2f ms%n", stats.getMedian());
        out.printf("90th Percentile: %.2f ms%n", stats.getPercentile90());
        out.printf("95th Percentile: %.2f ms%n", stats.getPercentile95());
        out.printf("99th Percentile: %.2f ms%n", stats.getPercentile99());
        out.printf("Min Response Time: %.2f ms%n", stats.getMin());
        out.printf("Max Response Time: %.2f ms%n", stats.getMax());
        out.printf("Standard Deviation: %.2f ms%n", stats.getStandardDeviation());
    }

    // Set the name that appears in JMeter's listener menu
    @Override
    public String getStaticLabel() {
        return "Combined Aggregate & Summary Report";
    }
}

Step 3: Understand the Underlying Data Collection #

The AbstractReport class uses a StatsHolder (from org.apache.jmeter.report.core) to accumulate all metrics needed by both reports:

• Total samples, error count, error percentage
• Average, min, max response times
• Key percentiles (50th, 90th, 95th, 99th)
• Throughput (requests per second)
• Standard deviation

You can access all these metrics directly via the inherited stats variable.

Step 4: Package and Deploy Your Listener #

1. Compile your Java class into a JAR file.
2. Copy the JAR into JMeter's lib/ext directory (found in your JMeter installation folder).
3. Restart JMeter.

Your new listener will now appear in the Listeners menu under the name you set in getStaticLabel().

Step 5: Customize Further (Optional) #

• Add a GUI: If you want a visual interface like the built-in reports, override methods like createPanel() to build a Swing UI. Reference the source code of AggregateReport or SummaryReport to see how their UIs are structured.
• Export to File: Modify printResults() to write data to CSV/JSON instead of the console.
• Filter Data: Add logic to exclude specific samples (e.g., successful vs. failed) from the stats.

Tips for Success #

• Check JMeter's Source: Look at the implementation of AggregateReport and SummaryReport in the Apache JMeter repo to understand how they handle data and UI.
• Test Thoroughly: Run a sample test plan and compare your listener's output to the built-in reports to ensure metrics match.
• Thread Safety: The StatsHolder from AbstractReport is already thread-safe, so you don't need to worry about concurrent test threads breaking your data collection.

内容的提问来源于stack exchange，提问作者Krishna Barri