Monitoring threadpool with gauge from dropWizard metric

Question

I want to use Guage metric from DropWizard to monitor my thread pool size.

        ThreadPoolExecutor threadPoolExecutor = new ThreadPoolExecutor(
            2,
            2,
            1,
            TimeUnit.MINUTES,
            new ArrayBlockingQueue<Runnable>(100),
            new ThreadPoolExecutor.DiscardPolicy());

    metricRegistry.register(name(ThreadPoolExecutor.class, "ThreadPoolRemainingCapacity"), (Gauge<Integer>) () -> threadPoolExecutor.getQueue().remainingCapacity());
    metricRegistry.register(name(ThreadPoolExecutor.class, "ThreadPoolOccupiedCapacity"), (Gauge<Integer>) () -> threadPoolExecutor.getQueue().size());

    return threadPoolExecutor;

From what I understand, this is done automatically in an interval time. But it seems that I don't get any monitoring data for my registered metrics. Although other metrics in my application such as counter and timer work just fine.

Can someone help me where I got it wrong?

Thank you.

where does your metric registery come from? How do you report on it? — pandaadb, Oct 31 '16 at 11:38

score 2 · Answer 1 · answered Feb 22 '17 at 14:38

You should register a class which implements the MetricSet interface like (this on has a few spring annotation you could ignore):

@Component
public class ThreadPoolGaugeSet implements MetricSet {

    @Autowired 
    @Qualifier( "taskExecutor")
    private ThreadPoolTaskExecutor taskExecutor;

    @Autowired 
    @Qualifier( "taskExecutorLowPrio")
    private ThreadPoolTaskExecutor lowPriotaskExecutor;

    @Override
    public Map<String, Metric> getMetrics() {
        final Map<String, Metric> gauges = new HashMap<>();

        gauges.put( "highPrioActive", (Gauge<Long>) () -> (long) taskExecutor.getActiveCount() );
        gauges.put( "lowPrioActive",  (Gauge<Long>) () -> (long) lowPriotaskExecutor.getActiveCount() );
        return gauges;
    }
}

...

metricRegistry.register( "threadpool", poolGauge );

score 0 · Answer 2 · answered Jun 26 '18 at 10:49

Your code should already be correct. You might do following to debug it:

Check whether your metrics appears at <admin_url>/metrics (if you're running on your machine it's default to localhost:8081/metrics). Usually there is some filtering if you send your metrics to 3rd party like prometheus or datadog.
Make sure to use the correct MetricRegistry instance for registering your metrics. Usually just use environment.metrics().
You might want to give a shorter name to your metrics temporarily to help debugging. The first parameter of register accepts plain string.

Monitoring threadpool with gauge from dropWizard metric

2 Answers2