ArrayIndexOutOfBoundsException when using LongTaskTimer

[rdehuyss]

Describe the bug / it may be a question
I'm seeing ArrayIndexOutOfBoundsException in my logs when using LongTaskTimer.

Environment

• Micrometer version 1.11.0
• Micrometer registry: SimpleMeterRegistry
• OS: macOS
• Java version: openjdk version "17.0.1" 2021-10-19

To Reproduce

    @Test
    void reproduceIssue() throws InterruptedException {
        SimpleMeterRegistry simpleMeterRegistry = new SimpleMeterRegistry();

        LongTaskTimer timer = LongTaskTimer.builder("jobrunr.jobs")
                .publishPercentiles(0.25, 0.5, 0.75, 0.8, 0.9, 0.95)
                .publishPercentileHistogram()
                .register(simpleMeterRegistry);
        LongTaskTimer.Sample start = timer.start();
        Thread.sleep(3640);
        System.out.println(simpleMeterRegistry.getMetersAsString());
        Thread.sleep(2640);
        System.out.println(start.stop());

        System.out.println(simpleMeterRegistry.getMetersAsString());
    }

This results in the following output:

08:46:23.327 [Test worker] WARN  i.m.c.i.internal.DefaultGauge - Failed to apply the value function for the gauge 'jobrunr.jobs.percentile'. Note that subsequent logs will be logged at debug level.
java.lang.ArrayIndexOutOfBoundsException: Index 5 out of bounds for length 5
	at io.micrometer.core.instrument.distribution.HistogramGauges.lambda$new$9(HistogramGauges.java:119)
	at io.micrometer.core.instrument.internal.DefaultGauge.value(DefaultGauge.java:53)
	at io.micrometer.core.instrument.Measurement.getValue(Measurement.java:54)
	at io.micrometer.core.instrument.simple.SimpleMeterRegistry.toString(SimpleMeterRegistry.java:206)
	at io.micrometer.core.instrument.simple.SimpleMeterRegistry.lambda$toString$1(SimpleMeterRegistry.java:195)
jobrunr.jobs(LONG_TASK_TIMER)[]; active_tasks=1.0, duration=3.651603291 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.25']; value=3.653785708 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.9']; value=3.653818208 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.8']; value=3.653818208 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.5']; value=3.653818208 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.95']; value=NaN seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.75']; value=3.653818208 seconds
6322804041
jobrunr.jobs(LONG_TASK_TIMER)[]; active_tasks=0.0, duration=0.0 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.25']; value=0.0 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.9']; value=0.0 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.8']; value=0.0 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.5']; value=0.0 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.95']; value=0.0 seconds
jobrunr.jobs.percentile(GAUGE)[phi='0.75']; value=0.0 seconds

Expected behavior
No ArrayIndexOutOfBoundsException should happen.

Additional context
NA

[jvwilge]

We're having a problem that looks similar and might be related :

Failed to apply the value function for the gauge 'http.server.requests.active.percentile'. Note that subsequent logs will be logged at debug level.
java.lang.ArrayIndexOutOfBoundsException: Index 3 out of bounds for length 3
	at io.micrometer.core.instrument.distribution.HistogramGauges.lambda$new$9(HistogramGauges.java:119)
	at io.micrometer.core.instrument.internal.DefaultGauge.value(DefaultGauge.java:53)
	at io.micrometer.signalfx.SignalFxMeterRegistry.addGauge(SignalFxMeterRegistry.java:227)
	at io.micrometer.core.instrument.Meter.match(Meter.java:100)
	at io.micrometer.signalfx.SignalFxMeterRegistry.lambda$publish$1(SignalFxMeterRegistry.java:117)
	at java.base/java.util.stream.ReferencePipeline$3$1.accept(Unknown Source)
	at java.base/java.util.ArrayList$SubList$2.forEachRemaining(Unknown Source)
	at java.base/java.util.stream.AbstractPipeline.copyInto(Unknown Source)
	at java.base/java.util.stream.AbstractPipeline.wrapAndCopyInto(Unknown Source)
	at java.base/java.util.stream.ForEachOps$ForEachOp.evaluateSequential(Unknown Source)
	at java.base/java.util.stream.ForEachOps$ForEachOp$OfRef.evaluateSequential(Unknown Source)
	at java.base/java.util.stream.AbstractPipeline.evaluate(Unknown Source)
	at java.base/java.util.stream.ReferencePipeline.forEach(Unknown Source)
	at io.micrometer.signalfx.SignalFxMeterRegistry.publish(SignalFxMeterRegistry.java:128)
	at io.micrometer.core.instrument.push.PushMeterRegistry.publishSafely(PushMeterRegistry.java:67)
	at java.base/java.util.concurrent.Executors$RunnableAdapter.call(Unknown Source)
	at java.base/java.util.concurrent.FutureTask.runAndReset(Unknown Source)
	at java.base/java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(Unknown Source)
	at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
	at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
	at java.base/java.lang.Thread.run(Unknown Source)

We're using SignalFxMeterRegistry and http.server.requests.active.percentile is something coming from the spring boot actuator.

When you change the example from @rdehuyss to .publishPercentiles(0.25, 0.51, 0.75, 0.8, 0.9, 0.95) it actually succeeds. So I suspect that

        List<Double> percentilesAboveInterpolatableLine = percentilesRequested.stream()
            .filter(p -> p * (activeTasks.size() + 1) > activeTasks.size())
            .collect(Collectors.toList());

in DefaultLongTaskTimer#takeSnapshot() might be the culprit (could it be > vs. >=?. It looks like one of the buckets is missing when creating the HistogramSnapshot in the return statement of the takeSnapshot() method.
The build of micrometer still succeeds after changing > into >=

There are also cases where multiple buckets are missing (0.11, 0.22, 0.33, 0.44) for example. When I add 3 samples it works again.