Java Code Examples for org.apache.flink.streaming.api.functions.source.FileProcessingMode#PROCESS_CONTINUOUSLY

The following examples show how to use org.apache.flink.streaming.api.functions.source.FileProcessingMode#PROCESS_CONTINUOUSLY . These examples are extracted from open source projects. You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may check out the related API usage on the sidebar.
Example 1
Source Project: flink   File: HiveTableSource.java    License: Apache License 2.0 5 votes vote down vote up
private DataStream<RowData> createStreamSourceForNonPartitionTable(
		StreamExecutionEnvironment execEnv,
		TypeInformation<RowData> typeInfo,
		HiveTableInputFormat inputFormat,
		HiveTablePartition hiveTable) {
	HiveTableFileInputFormat fileInputFormat = new HiveTableFileInputFormat(inputFormat, hiveTable);

	Configuration configuration = new Configuration();
	catalogTable.getOptions().forEach(configuration::setString);
	String consumeOrderStr = configuration.get(STREAMING_SOURCE_CONSUME_ORDER);
	ConsumeOrder consumeOrder = ConsumeOrder.getConsumeOrder(consumeOrderStr);
	if (consumeOrder != ConsumeOrder.CREATE_TIME_ORDER) {
		throw new UnsupportedOperationException(
				"Only " + ConsumeOrder.CREATE_TIME_ORDER + " is supported for non partition table.");
	}

	String consumeOffset = configuration.get(STREAMING_SOURCE_CONSUME_START_OFFSET);
	// to Local zone mills instead of UTC mills
	long currentReadTime = TimestampData.fromLocalDateTime(toLocalDateTime(consumeOffset))
			.toTimestamp().getTime();

	Duration monitorInterval = configuration.get(STREAMING_SOURCE_MONITOR_INTERVAL);

	ContinuousFileMonitoringFunction<RowData> monitoringFunction =
			new ContinuousFileMonitoringFunction<>(
					fileInputFormat,
					FileProcessingMode.PROCESS_CONTINUOUSLY,
					execEnv.getParallelism(),
					monitorInterval.toMillis(),
					currentReadTime);

	ContinuousFileReaderOperatorFactory<RowData, TimestampedFileInputSplit> factory =
			new ContinuousFileReaderOperatorFactory<>(fileInputFormat);

	String sourceName = "HiveFileMonitoringFunction";
	SingleOutputStreamOperator<RowData> source = execEnv.addSource(monitoringFunction, sourceName)
			.transform("Split Reader: " + sourceName, typeInfo, factory);

	return new DataStreamSource<>(source);
}
 
Example 2
@Test
public void testProgram() throws Exception {

	/*
	* This test checks the interplay between the monitor and the reader
	* and also the failExternally() functionality. To test the latter we
	* set the parallelism to 1 so that we have the chaining between the sink,
	* which throws the SuccessException to signal the end of the test, and the
	* reader.
	* */

	TextInputFormat format = new TextInputFormat(new Path(hdfsURI));
	format.setFilePath(hdfsURI);
	format.setFilesFilter(FilePathFilter.createDefaultFilter());

	// create the stream execution environment with a parallelism > 1 to test
	final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
	env.setParallelism(PARALLELISM);

	ContinuousFileMonitoringFunction<String> monitoringFunction =
		new ContinuousFileMonitoringFunction<>(format,
			FileProcessingMode.PROCESS_CONTINUOUSLY,
			env.getParallelism(), INTERVAL);

	// the monitor has always DOP 1
	DataStream<TimestampedFileInputSplit> splits = env.addSource(monitoringFunction);
	Assert.assertEquals(1, splits.getParallelism());

	ContinuousFileReaderOperator<String> reader = new ContinuousFileReaderOperator<>(format);
	TypeInformation<String> typeInfo = TypeExtractor.getInputFormatTypes(format);

	// the readers can be multiple
	DataStream<String> content = splits.transform("FileSplitReader", typeInfo, reader);
	Assert.assertEquals(PARALLELISM, content.getParallelism());

	// finally for the sink we set the parallelism to 1 so that we can verify the output
	TestingSinkFunction sink = new TestingSinkFunction();
	content.addSink(sink).setParallelism(1);

	Thread job = new Thread() {

		@Override
		public void run() {
			try {
				env.execute("ContinuousFileProcessingITCase Job.");
			} catch (Exception e) {
				Throwable th = e;
				for (int depth = 0; depth < 20; depth++) {
					if (th instanceof SuccessException) {
						return;
					} else if (th.getCause() != null) {
						th = th.getCause();
					} else {
						break;
					}
				}
				e.printStackTrace();
				Assert.fail(e.getMessage());
			}
		}
	};
	job.start();

	// The modification time of the last created file.
	long lastCreatedModTime = Long.MIN_VALUE;

	// create the files to be read
	for (int i = 0; i < NO_OF_FILES; i++) {
		Tuple2<org.apache.hadoop.fs.Path, String> tmpFile;
		long modTime;
		do {

			// give it some time so that the files have
			// different modification timestamps.
			Thread.sleep(50);

			tmpFile = fillWithData(hdfsURI, "file", i, "This is test line.");

			modTime = hdfs.getFileStatus(tmpFile.f0).getModificationTime();
			if (modTime <= lastCreatedModTime) {
				// delete the last created file to recreate it with a different timestamp
				hdfs.delete(tmpFile.f0, false);
			}
		} while (modTime <= lastCreatedModTime);
		lastCreatedModTime = modTime;

		// put the contents in the expected results list before the reader picks them
		// this is to guarantee that they are in before the reader finishes (avoid race conditions)
		expectedContents.put(i, tmpFile.f1);

		org.apache.hadoop.fs.Path file =
			new org.apache.hadoop.fs.Path(hdfsURI + "/file" + i);
		hdfs.rename(tmpFile.f0, file);
		Assert.assertTrue(hdfs.exists(file));
	}

	// wait for the job to finish.
	job.join();
}
 
Example 3
/**
 * Manually run this to write binary snapshot data. Remove @Ignore to run.
 */
@Ignore
@Test
public void writeMonitoringSourceSnapshot() throws Exception {

	File testFolder = tempFolder.newFolder();

	long fileModTime = Long.MIN_VALUE;
	for (int i = 0; i < 1; i++) {
		Tuple2<File, String> file = createFileAndFillWithData(testFolder, "file", i, "This is test line.");
		fileModTime = file.f0.lastModified();
	}

	TextInputFormat format = new TextInputFormat(new Path(testFolder.getAbsolutePath()));

	final ContinuousFileMonitoringFunction<String> monitoringFunction =
		new ContinuousFileMonitoringFunction<>(format, FileProcessingMode.PROCESS_CONTINUOUSLY, 1, INTERVAL);

	StreamSource<TimestampedFileInputSplit, ContinuousFileMonitoringFunction<String>> src =
		new StreamSource<>(monitoringFunction);

	final AbstractStreamOperatorTestHarness<TimestampedFileInputSplit> testHarness =
			new AbstractStreamOperatorTestHarness<>(src, 1, 1, 0);

	testHarness.open();

	final Throwable[] error = new Throwable[1];

	final OneShotLatch latch = new OneShotLatch();

	// run the source asynchronously
	Thread runner = new Thread() {
		@Override
		public void run() {
			try {
				monitoringFunction.run(new DummySourceContext() {
					@Override
					public void collect(TimestampedFileInputSplit element) {
						latch.trigger();
					}

					@Override
					public void markAsTemporarilyIdle() {

					}
				});
			}
			catch (Throwable t) {
				t.printStackTrace();
				error[0] = t;
			}
		}
	};
	runner.start();

	if (!latch.isTriggered()) {
		latch.await();
	}

	final OperatorSubtaskState snapshot;
	synchronized (testHarness.getCheckpointLock()) {
		snapshot = testHarness.snapshot(0L, 0L);
	}

	OperatorSnapshotUtil.writeStateHandle(
			snapshot,
			"src/test/resources/monitoring-function-migration-test-" + fileModTime + "-flink" + flinkGenerateSavepointVersion + "-snapshot");

	monitoringFunction.cancel();
	runner.join();

	testHarness.close();
}
 
Example 4
@Test
public void testProgram() throws Exception {

	/*
	* This test checks the interplay between the monitor and the reader
	* and also the failExternally() functionality. To test the latter we
	* set the parallelism to 1 so that we have the chaining between the sink,
	* which throws the SuccessException to signal the end of the test, and the
	* reader.
	* */

	TextInputFormat format = new TextInputFormat(new Path(hdfsURI));
	format.setFilePath(hdfsURI);
	format.setFilesFilter(FilePathFilter.createDefaultFilter());

	// create the stream execution environment with a parallelism > 1 to test
	final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
	env.setParallelism(PARALLELISM);

	ContinuousFileMonitoringFunction<String> monitoringFunction =
		new ContinuousFileMonitoringFunction<>(format,
			FileProcessingMode.PROCESS_CONTINUOUSLY,
			env.getParallelism(), INTERVAL);

	// the monitor has always DOP 1
	DataStream<TimestampedFileInputSplit> splits = env.addSource(monitoringFunction);
	Assert.assertEquals(1, splits.getParallelism());

	ContinuousFileReaderOperator<String> reader = new ContinuousFileReaderOperator<>(format);
	TypeInformation<String> typeInfo = TypeExtractor.getInputFormatTypes(format);

	// the readers can be multiple
	DataStream<String> content = splits.transform("FileSplitReader", typeInfo, reader);
	Assert.assertEquals(PARALLELISM, content.getParallelism());

	// finally for the sink we set the parallelism to 1 so that we can verify the output
	TestingSinkFunction sink = new TestingSinkFunction();
	content.addSink(sink).setParallelism(1);

	Thread job = new Thread() {

		@Override
		public void run() {
			try {
				env.execute("ContinuousFileProcessingITCase Job.");
			} catch (Exception e) {
				Throwable th = e;
				for (int depth = 0; depth < 20; depth++) {
					if (th instanceof SuccessException) {
						return;
					} else if (th.getCause() != null) {
						th = th.getCause();
					} else {
						break;
					}
				}
				e.printStackTrace();
				Assert.fail(e.getMessage());
			}
		}
	};
	job.start();

	// The modification time of the last created file.
	long lastCreatedModTime = Long.MIN_VALUE;

	// create the files to be read
	for (int i = 0; i < NO_OF_FILES; i++) {
		Tuple2<org.apache.hadoop.fs.Path, String> tmpFile;
		long modTime;
		do {

			// give it some time so that the files have
			// different modification timestamps.
			Thread.sleep(50);

			tmpFile = fillWithData(hdfsURI, "file", i, "This is test line.");

			modTime = hdfs.getFileStatus(tmpFile.f0).getModificationTime();
			if (modTime <= lastCreatedModTime) {
				// delete the last created file to recreate it with a different timestamp
				hdfs.delete(tmpFile.f0, false);
			}
		} while (modTime <= lastCreatedModTime);
		lastCreatedModTime = modTime;

		// put the contents in the expected results list before the reader picks them
		// this is to guarantee that they are in before the reader finishes (avoid race conditions)
		expectedContents.put(i, tmpFile.f1);

		org.apache.hadoop.fs.Path file =
			new org.apache.hadoop.fs.Path(hdfsURI + "/file" + i);
		hdfs.rename(tmpFile.f0, file);
		Assert.assertTrue(hdfs.exists(file));
	}

	// wait for the job to finish.
	job.join();
}
 
Example 5
/**
 * Manually run this to write binary snapshot data. Remove @Ignore to run.
 */
@Ignore
@Test
public void writeMonitoringSourceSnapshot() throws Exception {

	File testFolder = tempFolder.newFolder();

	long fileModTime = Long.MIN_VALUE;
	for (int i = 0; i < 1; i++) {
		Tuple2<File, String> file = createFileAndFillWithData(testFolder, "file", i, "This is test line.");
		fileModTime = file.f0.lastModified();
	}

	TextInputFormat format = new TextInputFormat(new Path(testFolder.getAbsolutePath()));

	final ContinuousFileMonitoringFunction<String> monitoringFunction =
		new ContinuousFileMonitoringFunction<>(format, FileProcessingMode.PROCESS_CONTINUOUSLY, 1, INTERVAL);

	StreamSource<TimestampedFileInputSplit, ContinuousFileMonitoringFunction<String>> src =
		new StreamSource<>(monitoringFunction);

	final AbstractStreamOperatorTestHarness<TimestampedFileInputSplit> testHarness =
			new AbstractStreamOperatorTestHarness<>(src, 1, 1, 0);

	testHarness.open();

	final Throwable[] error = new Throwable[1];

	final OneShotLatch latch = new OneShotLatch();

	// run the source asynchronously
	Thread runner = new Thread() {
		@Override
		public void run() {
			try {
				monitoringFunction.run(new DummySourceContext() {
					@Override
					public void collect(TimestampedFileInputSplit element) {
						latch.trigger();
					}

					@Override
					public void markAsTemporarilyIdle() {

					}
				});
			}
			catch (Throwable t) {
				t.printStackTrace();
				error[0] = t;
			}
		}
	};
	runner.start();

	if (!latch.isTriggered()) {
		latch.await();
	}

	final OperatorSubtaskState snapshot;
	synchronized (testHarness.getCheckpointLock()) {
		snapshot = testHarness.snapshot(0L, 0L);
	}

	OperatorSnapshotUtil.writeStateHandle(
			snapshot,
			"src/test/resources/monitoring-function-migration-test-" + fileModTime + "-flink" + flinkGenerateSavepointVersion + "-snapshot");

	monitoringFunction.cancel();
	runner.join();

	testHarness.close();
}
 
Example 6
@Test
public void testProgram() throws Exception {

	/*
	* This test checks the interplay between the monitor and the reader
	* and also the failExternally() functionality. To test the latter we
	* set the parallelism to 1 so that we have the chaining between the sink,
	* which throws the SuccessException to signal the end of the test, and the
	* reader.
	* */

	TextInputFormat format = new TextInputFormat(new Path(hdfsURI));
	format.setFilePath(hdfsURI);
	format.setFilesFilter(FilePathFilter.createDefaultFilter());

	// create the stream execution environment with a parallelism > 1 to test
	final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
	env.setParallelism(PARALLELISM);

	ContinuousFileMonitoringFunction<String> monitoringFunction =
		new ContinuousFileMonitoringFunction<>(format,
			FileProcessingMode.PROCESS_CONTINUOUSLY,
			env.getParallelism(), INTERVAL);

	// the monitor has always DOP 1
	DataStream<TimestampedFileInputSplit> splits = env.addSource(monitoringFunction);
	Assert.assertEquals(1, splits.getParallelism());

	TypeInformation<String> typeInfo = TypeExtractor.getInputFormatTypes(format);

	// the readers can be multiple
	DataStream<String> content = splits.transform("FileSplitReader", typeInfo, new ContinuousFileReaderOperatorFactory<>(format));
	Assert.assertEquals(PARALLELISM, content.getParallelism());

	// finally for the sink we set the parallelism to 1 so that we can verify the output
	TestingSinkFunction sink = new TestingSinkFunction();
	content.addSink(sink).setParallelism(1);

	CompletableFuture<Void> jobFuture = new CompletableFuture<>();
	new Thread(() -> {
		try {
			env.execute("ContinuousFileProcessingITCase Job.");
			jobFuture.complete(null);
		} catch (Exception e) {
			if (ExceptionUtils.findThrowable(e, SuccessException.class).isPresent()) {
				jobFuture.complete(null);
			} else {
				jobFuture.completeExceptionally(e);
			}
		}
	}).start();

	// The modification time of the last created file.
	long lastCreatedModTime = Long.MIN_VALUE;

	// create the files to be read
	for (int i = 0; i < NO_OF_FILES; i++) {
		Tuple2<org.apache.hadoop.fs.Path, String> tmpFile;
		long modTime;
		do {

			// give it some time so that the files have
			// different modification timestamps.
			Thread.sleep(50);

			tmpFile = fillWithData(hdfsURI, "file", i, "This is test line.");

			modTime = hdfs.getFileStatus(tmpFile.f0).getModificationTime();
			if (modTime <= lastCreatedModTime) {
				// delete the last created file to recreate it with a different timestamp
				hdfs.delete(tmpFile.f0, false);
			}
		} while (modTime <= lastCreatedModTime);
		lastCreatedModTime = modTime;

		// put the contents in the expected results list before the reader picks them
		// this is to guarantee that they are in before the reader finishes (avoid race conditions)
		expectedContents.put(i, tmpFile.f1);

		org.apache.hadoop.fs.Path file =
			new org.apache.hadoop.fs.Path(hdfsURI + "/file" + i);
		hdfs.rename(tmpFile.f0, file);
		Assert.assertTrue(hdfs.exists(file));
	}

	jobFuture.get();
}
 
Example 7
/**
 * Manually run this to write binary snapshot data. Remove @Ignore to run.
 */
@Ignore
@Test
public void writeMonitoringSourceSnapshot() throws Exception {

	File testFolder = tempFolder.newFolder();

	long fileModTime = Long.MIN_VALUE;
	for (int i = 0; i < 1; i++) {
		Tuple2<File, String> file = createFileAndFillWithData(testFolder, "file", i, "This is test line.");
		fileModTime = file.f0.lastModified();
	}

	TextInputFormat format = new TextInputFormat(new Path(testFolder.getAbsolutePath()));

	final ContinuousFileMonitoringFunction<String> monitoringFunction =
		new ContinuousFileMonitoringFunction<>(format, FileProcessingMode.PROCESS_CONTINUOUSLY, 1, INTERVAL);

	StreamSource<TimestampedFileInputSplit, ContinuousFileMonitoringFunction<String>> src =
		new StreamSource<>(monitoringFunction);

	final AbstractStreamOperatorTestHarness<TimestampedFileInputSplit> testHarness =
			new AbstractStreamOperatorTestHarness<>(src, 1, 1, 0);

	testHarness.open();

	final Throwable[] error = new Throwable[1];

	final OneShotLatch latch = new OneShotLatch();

	// run the source asynchronously
	Thread runner = new Thread() {
		@Override
		public void run() {
			try {
				monitoringFunction.run(new DummySourceContext() {
					@Override
					public void collect(TimestampedFileInputSplit element) {
						latch.trigger();
					}

					@Override
					public void markAsTemporarilyIdle() {

					}
				});
			}
			catch (Throwable t) {
				t.printStackTrace();
				error[0] = t;
			}
		}
	};
	runner.start();

	if (!latch.isTriggered()) {
		latch.await();
	}

	final OperatorSubtaskState snapshot;
	synchronized (testHarness.getCheckpointLock()) {
		snapshot = testHarness.snapshot(0L, 0L);
	}

	OperatorSnapshotUtil.writeStateHandle(
			snapshot,
			"src/test/resources/monitoring-function-migration-test-" + fileModTime + "-flink" + flinkGenerateSavepointVersion + "-snapshot");

	monitoringFunction.cancel();
	runner.join();

	testHarness.close();
}
 
Example 8
@Test
public void testMonitoringSourceRestore() throws Exception {

	File testFolder = tempFolder.newFolder();

	TextInputFormat format = new TextInputFormat(new Path(testFolder.getAbsolutePath()));

	final ContinuousFileMonitoringFunction<String> monitoringFunction =
		new ContinuousFileMonitoringFunction<>(format, FileProcessingMode.PROCESS_CONTINUOUSLY, 1, INTERVAL);

	StreamSource<TimestampedFileInputSplit, ContinuousFileMonitoringFunction<String>> src =
		new StreamSource<>(monitoringFunction);

	final AbstractStreamOperatorTestHarness<TimestampedFileInputSplit> testHarness =
		new AbstractStreamOperatorTestHarness<>(src, 1, 1, 0);

	testHarness.setup();

	testHarness.initializeState(
		OperatorSnapshotUtil.getResourceFilename(
			"monitoring-function-migration-test-" + expectedModTime + "-flink" + testMigrateVersion + "-snapshot"));

	testHarness.open();

	Assert.assertEquals((long) expectedModTime, monitoringFunction.getGlobalModificationTime());

}
 
Example 9
@Test
public void testMonitoringSourceRestore() throws Exception {

	File testFolder = tempFolder.newFolder();

	TextInputFormat format = new TextInputFormat(new Path(testFolder.getAbsolutePath()));

	final ContinuousFileMonitoringFunction<String> monitoringFunction =
		new ContinuousFileMonitoringFunction<>(format, FileProcessingMode.PROCESS_CONTINUOUSLY, 1, INTERVAL);

	StreamSource<TimestampedFileInputSplit, ContinuousFileMonitoringFunction<String>> src =
		new StreamSource<>(monitoringFunction);

	final AbstractStreamOperatorTestHarness<TimestampedFileInputSplit> testHarness =
		new AbstractStreamOperatorTestHarness<>(src, 1, 1, 0);

	testHarness.setup();

	testHarness.initializeState(
		OperatorSnapshotUtil.getResourceFilename(
			"monitoring-function-migration-test-" + expectedModTime + "-flink" + testMigrateVersion + "-snapshot"));

	testHarness.open();

	Assert.assertEquals((long) expectedModTime, monitoringFunction.getGlobalModificationTime());

}
 
Example 10
@Test
public void testMonitoringSourceRestore() throws Exception {

	File testFolder = tempFolder.newFolder();

	TextInputFormat format = new TextInputFormat(new Path(testFolder.getAbsolutePath()));

	final ContinuousFileMonitoringFunction<String> monitoringFunction =
		new ContinuousFileMonitoringFunction<>(format, FileProcessingMode.PROCESS_CONTINUOUSLY, 1, INTERVAL);

	StreamSource<TimestampedFileInputSplit, ContinuousFileMonitoringFunction<String>> src =
		new StreamSource<>(monitoringFunction);

	final AbstractStreamOperatorTestHarness<TimestampedFileInputSplit> testHarness =
		new AbstractStreamOperatorTestHarness<>(src, 1, 1, 0);

	testHarness.setup();

	testHarness.initializeState(
		OperatorSnapshotUtil.getResourceFilename(
			"monitoring-function-migration-test-" + expectedModTime + "-flink" + testMigrateVersion + "-snapshot"));

	testHarness.open();

	Assert.assertEquals((long) expectedModTime, monitoringFunction.getGlobalModificationTime());

}