Why is the monitoring interface time (TOTAL_TIME) so large when using Spring Boot Actuator to monitor the API of a Java web project?

Question

A Spring Boot project has an interface, and the Actuator is used to monitor the number of calls and time for that interface, using the @Timed annotation on a certain request method.

After completing tests with JMeter, where the testing process is far less than 10 minutes, accessing http://xxx:xxx/actuator/metrics/METRICS_NAME shows that the TOTAL_TIME parameter is around 2000 seconds, with a COUNT of 40,000.

The question is: TOTAL_TIME is over 2000 seconds, about 33 minutes, which is much greater than the test process time. Is this normal?

When I use JMeter for testing, with a small number of threads and loop counts, the time obtained is not particularly large. However, when the number of threads or loop counts is increased, the final value of TOTAL_TIME becomes very large.

Answer 1

TOTAL_TIME is sum of all response times since application start (or metrics reset)

Instead of comparing it to test duration you should compare it to the sum of elapsed times of all HTTP Request samplers which are in your .jtl results file

If you want you can calculate the sum of all Samplers response times somewhere in tearDown Thread Group using JSR223 Sampler and the following Groovy code:

def total_time = 0

new File('/path/to/result/file.jtl').withReader { reader ->
    reader.eachLine { line, lineNumber ->
        if (lineNumber > 1) {
            def columns = line.split(',')
            def elapsed = columns[1].toBigDecimal()
            if (elapsed) {
                total_time += elapsed
            }
        }
    }
}