use of org.apache.druid.testing.utils.JsonEventSerializer in project druid by druid-io.
the class AbstractStreamIndexingTest method testIndexWithLosingNodeHelper.
private void testIndexWithLosingNodeHelper(Runnable restartRunnable, Runnable waitForReadyRunnable, @Nullable Boolean transactionEnabled) throws Exception {
final GeneratedTestConfig generatedTestConfig = new GeneratedTestConfig(INPUT_FORMAT, getResourceAsString(JSON_INPUT_FORMAT_PATH));
try (final Closeable closer = createResourceCloser(generatedTestConfig);
final StreamEventWriter streamEventWriter = createStreamEventWriter(config, transactionEnabled)) {
final String taskSpec = generatedTestConfig.getStreamIngestionPropsTransform().apply(getResourceAsString(SUPERVISOR_SPEC_TEMPLATE_PATH));
LOG.info("supervisorSpec: [%s]\n", taskSpec);
// Start supervisor
generatedTestConfig.setSupervisorId(indexer.submitSupervisor(taskSpec));
LOG.info("Submitted supervisor");
// Start generating one third of the data (before restarting)
int secondsToGenerateRemaining = TOTAL_NUMBER_OF_SECOND;
int secondsToGenerateFirstRound = TOTAL_NUMBER_OF_SECOND / 3;
secondsToGenerateRemaining = secondsToGenerateRemaining - secondsToGenerateFirstRound;
final StreamGenerator streamGenerator = new WikipediaStreamEventStreamGenerator(new JsonEventSerializer(jsonMapper), EVENTS_PER_SECOND, CYCLE_PADDING_MS);
long numWritten = streamGenerator.run(generatedTestConfig.getStreamName(), streamEventWriter, secondsToGenerateFirstRound, FIRST_EVENT_TIME);
// Verify supervisor is healthy before restart
ITRetryUtil.retryUntil(() -> SupervisorStateManager.BasicState.RUNNING.equals(indexer.getSupervisorStatus(generatedTestConfig.getSupervisorId())), true, 10000, 30, "Waiting for supervisor to be healthy");
// Restart Druid process
LOG.info("Restarting Druid process");
restartRunnable.run();
LOG.info("Restarted Druid process");
// Start generating one third of the data (while restarting)
int secondsToGenerateSecondRound = TOTAL_NUMBER_OF_SECOND / 3;
secondsToGenerateRemaining = secondsToGenerateRemaining - secondsToGenerateSecondRound;
numWritten += streamGenerator.run(generatedTestConfig.getStreamName(), streamEventWriter, secondsToGenerateSecondRound, FIRST_EVENT_TIME.plusSeconds(secondsToGenerateFirstRound));
// Wait for Druid process to be available
LOG.info("Waiting for Druid process to be available");
waitForReadyRunnable.run();
LOG.info("Druid process is now available");
// Start generating remaining data (after restarting)
numWritten += streamGenerator.run(generatedTestConfig.getStreamName(), streamEventWriter, secondsToGenerateRemaining, FIRST_EVENT_TIME.plusSeconds(secondsToGenerateFirstRound + secondsToGenerateSecondRound));
// Verify supervisor is healthy
ITRetryUtil.retryUntil(() -> SupervisorStateManager.BasicState.RUNNING.equals(indexer.getSupervisorStatus(generatedTestConfig.getSupervisorId())), true, 10000, 30, "Waiting for supervisor to be healthy");
// Verify that supervisor ingested all data
verifyIngestedData(generatedTestConfig, numWritten);
}
}
use of org.apache.druid.testing.utils.JsonEventSerializer in project druid by druid-io.
the class AbstractStreamIndexingTest method testIndexWithStreamReshardHelper.
private void testIndexWithStreamReshardHelper(@Nullable Boolean transactionEnabled, int newShardCount) throws Exception {
final GeneratedTestConfig generatedTestConfig = new GeneratedTestConfig(INPUT_FORMAT, getResourceAsString(JSON_INPUT_FORMAT_PATH));
try (final Closeable closer = createResourceCloser(generatedTestConfig);
final StreamEventWriter streamEventWriter = createStreamEventWriter(config, transactionEnabled)) {
final String taskSpec = generatedTestConfig.getStreamIngestionPropsTransform().apply(getResourceAsString(SUPERVISOR_SPEC_TEMPLATE_PATH));
LOG.info("supervisorSpec: [%s]\n", taskSpec);
// Start supervisor
generatedTestConfig.setSupervisorId(indexer.submitSupervisor(taskSpec));
LOG.info("Submitted supervisor");
// Start generating one third of the data (before resharding)
int secondsToGenerateRemaining = TOTAL_NUMBER_OF_SECOND;
int secondsToGenerateFirstRound = TOTAL_NUMBER_OF_SECOND / 3;
secondsToGenerateRemaining = secondsToGenerateRemaining - secondsToGenerateFirstRound;
final StreamGenerator streamGenerator = new WikipediaStreamEventStreamGenerator(new JsonEventSerializer(jsonMapper), EVENTS_PER_SECOND, CYCLE_PADDING_MS);
long numWritten = streamGenerator.run(generatedTestConfig.getStreamName(), streamEventWriter, secondsToGenerateFirstRound, FIRST_EVENT_TIME);
// Verify supervisor is healthy before resahrding
ITRetryUtil.retryUntil(() -> SupervisorStateManager.BasicState.RUNNING.equals(indexer.getSupervisorStatus(generatedTestConfig.getSupervisorId())), true, 10000, 30, "Waiting for supervisor to be healthy");
// Reshard the supervisor by split from STREAM_SHARD_COUNT to newShardCount and waits until the resharding starts
streamAdminClient.updatePartitionCount(generatedTestConfig.getStreamName(), newShardCount, true);
// Start generating one third of the data (while resharding)
int secondsToGenerateSecondRound = TOTAL_NUMBER_OF_SECOND / 3;
secondsToGenerateRemaining = secondsToGenerateRemaining - secondsToGenerateSecondRound;
numWritten += streamGenerator.run(generatedTestConfig.getStreamName(), streamEventWriter, secondsToGenerateSecondRound, FIRST_EVENT_TIME.plusSeconds(secondsToGenerateFirstRound));
// Wait for stream to finish resharding
ITRetryUtil.retryUntil(() -> streamAdminClient.isStreamActive(generatedTestConfig.getStreamName()), true, 10000, 30, "Waiting for stream to finish resharding");
ITRetryUtil.retryUntil(() -> streamAdminClient.verfiyPartitionCountUpdated(generatedTestConfig.getStreamName(), STREAM_SHARD_COUNT, newShardCount), true, 10000, 30, "Waiting for stream to finish resharding");
// Start generating remaining data (after resharding)
numWritten += streamGenerator.run(generatedTestConfig.getStreamName(), streamEventWriter, secondsToGenerateRemaining, FIRST_EVENT_TIME.plusSeconds(secondsToGenerateFirstRound + secondsToGenerateSecondRound));
// Verify supervisor is healthy after resahrding
ITRetryUtil.retryUntil(() -> SupervisorStateManager.BasicState.RUNNING.equals(indexer.getSupervisorStatus(generatedTestConfig.getSupervisorId())), true, 10000, 30, "Waiting for supervisor to be healthy");
// Verify that supervisor can catch up with the stream
verifyIngestedData(generatedTestConfig, numWritten);
}
// Verify that event thrown away count was not incremented by the reshard
List<TaskResponseObject> completedTasks = indexer.getCompleteTasksForDataSource(generatedTestConfig.getFullDatasourceName());
for (TaskResponseObject task : completedTasks) {
try {
RowIngestionMetersTotals stats = indexer.getTaskStats(task.getId());
Assert.assertEquals(0L, stats.getThrownAway());
} catch (Exception e) {
// Failed task may not have a task stats report. We can ignore it as the task did not consume any data
if (!task.getStatus().isFailure()) {
throw e;
}
}
}
}
use of org.apache.druid.testing.utils.JsonEventSerializer in project druid by druid-io.
the class AbstractStreamIndexingTest method doTestIndexDataWithStartStopSupervisor.
protected void doTestIndexDataWithStartStopSupervisor(@Nullable Boolean transactionEnabled) throws Exception {
final GeneratedTestConfig generatedTestConfig = new GeneratedTestConfig(INPUT_FORMAT, getResourceAsString(JSON_INPUT_FORMAT_PATH));
try (final Closeable closer = createResourceCloser(generatedTestConfig);
final StreamEventWriter streamEventWriter = createStreamEventWriter(config, transactionEnabled)) {
final String taskSpec = generatedTestConfig.getStreamIngestionPropsTransform().apply(getResourceAsString(SUPERVISOR_SPEC_TEMPLATE_PATH));
LOG.info("supervisorSpec: [%s]\n", taskSpec);
// Start supervisor
generatedTestConfig.setSupervisorId(indexer.submitSupervisor(taskSpec));
LOG.info("Submitted supervisor");
// Start generating half of the data
int secondsToGenerateRemaining = TOTAL_NUMBER_OF_SECOND;
int secondsToGenerateFirstRound = TOTAL_NUMBER_OF_SECOND / 2;
secondsToGenerateRemaining = secondsToGenerateRemaining - secondsToGenerateFirstRound;
final StreamGenerator streamGenerator = new WikipediaStreamEventStreamGenerator(new JsonEventSerializer(jsonMapper), EVENTS_PER_SECOND, CYCLE_PADDING_MS);
long numWritten = streamGenerator.run(generatedTestConfig.getStreamName(), streamEventWriter, secondsToGenerateFirstRound, FIRST_EVENT_TIME);
// Verify supervisor is healthy before suspension
ITRetryUtil.retryUntil(() -> SupervisorStateManager.BasicState.RUNNING.equals(indexer.getSupervisorStatus(generatedTestConfig.getSupervisorId())), true, 10000, 30, "Waiting for supervisor to be healthy");
// Suspend the supervisor
indexer.suspendSupervisor(generatedTestConfig.getSupervisorId());
// Start generating remainning half of the data
numWritten += streamGenerator.run(generatedTestConfig.getStreamName(), streamEventWriter, secondsToGenerateRemaining, FIRST_EVENT_TIME.plusSeconds(secondsToGenerateFirstRound));
// Resume the supervisor
indexer.resumeSupervisor(generatedTestConfig.getSupervisorId());
// Verify supervisor is healthy after suspension
ITRetryUtil.retryUntil(() -> SupervisorStateManager.BasicState.RUNNING.equals(indexer.getSupervisorStatus(generatedTestConfig.getSupervisorId())), true, 10000, 30, "Waiting for supervisor to be healthy");
// Verify that supervisor can catch up with the stream
verifyIngestedData(generatedTestConfig, numWritten);
}
}
use of org.apache.druid.testing.utils.JsonEventSerializer in project druid by druid-io.
the class AbstractStreamIndexingTest method doTestIndexDataWithAutoscaler.
protected void doTestIndexDataWithAutoscaler(@Nullable Boolean transactionEnabled) throws Exception {
final GeneratedTestConfig generatedTestConfig = new GeneratedTestConfig(INPUT_FORMAT, getResourceAsString(JSON_INPUT_FORMAT_PATH));
try (final Closeable closer = createResourceCloser(generatedTestConfig);
final StreamEventWriter streamEventWriter = createStreamEventWriter(config, transactionEnabled)) {
final String taskSpec = generatedTestConfig.getStreamIngestionPropsTransform().apply(getResourceAsString(SUPERVISOR_WITH_AUTOSCALER_SPEC_TEMPLATE_PATH));
LOG.info("supervisorSpec: [%s]\n", taskSpec);
// Start supervisor
generatedTestConfig.setSupervisorId(indexer.submitSupervisor(taskSpec));
LOG.info("Submitted supervisor");
// Start generating half of the data
int secondsToGenerateRemaining = TOTAL_NUMBER_OF_SECOND;
int secondsToGenerateFirstRound = TOTAL_NUMBER_OF_SECOND / 2;
secondsToGenerateRemaining = secondsToGenerateRemaining - secondsToGenerateFirstRound;
final StreamGenerator streamGenerator = new WikipediaStreamEventStreamGenerator(new JsonEventSerializer(jsonMapper), EVENTS_PER_SECOND, CYCLE_PADDING_MS);
long numWritten = streamGenerator.run(generatedTestConfig.getStreamName(), streamEventWriter, secondsToGenerateFirstRound, FIRST_EVENT_TIME);
// Verify supervisor is healthy before suspension
ITRetryUtil.retryUntil(() -> SupervisorStateManager.BasicState.RUNNING.equals(indexer.getSupervisorStatus(generatedTestConfig.getSupervisorId())), true, 10000, 30, "Waiting for supervisor to be healthy");
// wait for autoScaling task numbers from 1 to 2.
ITRetryUtil.retryUntil(() -> indexer.getRunningTasks().size() == 2, true, 10000, 50, "waiting for autoScaling task numbers from 1 to 2");
// Start generating remainning half of the data
numWritten += streamGenerator.run(generatedTestConfig.getStreamName(), streamEventWriter, secondsToGenerateRemaining, FIRST_EVENT_TIME.plusSeconds(secondsToGenerateFirstRound));
// Verify that supervisor can catch up with the stream
verifyIngestedData(generatedTestConfig, numWritten);
}
}
Aggregations