kamalcph commented on code in PR #13969: URL: https://github.com/apache/kafka/pull/13969#discussion_r1257268739
########## clients/src/test/java/org/apache/kafka/common/metrics/SensorTest.java: ########## @@ -366,4 +368,60 @@ public void testUpdatingMetricConfigIsReflectedInTheSensor() { metrics.close(); } + + @Test + public void testConcurrentReadWriteAccessForMetrics() throws ExecutionException, InterruptedException { + final Metrics metrics = new Metrics(new MetricConfig().quota(Quota.upperBound(Double.MAX_VALUE)) + .timeWindow(1, TimeUnit.MILLISECONDS) + .samples(100)); + final Sensor sensor = metrics.sensor("sensor"); + + final int metricCount = 10000; + // Add a large number of metrics to increase the execution time of checkQuotas + for (int i = 0; i < metricCount; i++) { + sensor.add(metrics.metricName("test-metric" + i, "test-group" + i), new Rate()); + } + sensor.record(10, 1, false); + CountDownLatch latch = new CountDownLatch(1); + ExecutorService executor = Executors.newFixedThreadPool(2); + + // Use non-thread-safe methods for concurrent read and write of metrics. + Future<Throwable> worker = executor.submit(() -> { + try { + sensor.checkQuotasNonThreadSafe(3); + return null; + } catch (Throwable e) { + return e; + } + }); + executor.submit(() -> { + sensor.add(metrics.metricName("test-metric-non-thread-safe", "test-group-non-thread-safe"), new Rate()); + latch.countDown(); + }); + + assertTrue(latch.await(5, TimeUnit.SECONDS), "If this failure happen frequently, we can try to increase the wait time"); + assertTrue(worker.isDone()); + assertNotNull(worker.get()); + assertEquals(ConcurrentModificationException.class, worker.get().getClass()); Review Comment: Will the test throw exceptions consistently? What if `sensor.checkQuotasNonThreadSafe(3)` ran before we are adding a new metric? ########## clients/src/main/java/org/apache/kafka/common/metrics/Sensor.java: ########## @@ -250,7 +250,16 @@ public void checkQuotas() { checkQuotas(time.milliseconds()); } - public void checkQuotas(long timeMs) { + public synchronized void checkQuotas(long timeMs) { + checkQuotasInternal(timeMs); + } + + // visible for testing + protected void checkQuotasNonThreadSafe(long timeMs) { Review Comment: Can we make it package-private instead of protected? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: jira-unsubscr...@kafka.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org