Examples with FSDataOutputStream org.apache.flink.core.fs.FSDataOutputStream used on opensource projects

Example 16 with FSDataOutputStream

use of org.apache.flink.core.fs.FSDataOutputStream in project flink by apache.

the class SavepointStoreTest method testUnexpectedSavepoint.

/**
	 * Tests loading with unexpected magic number.
	 */
@Test
public void testUnexpectedSavepoint() throws Exception {
    // Random file
    Path filePath = new Path(tmp.getRoot().getPath(), UUID.randomUUID().toString());
    FSDataOutputStream fdos = FileSystem.get(filePath.toUri()).create(filePath, false);
    DataOutputStream dos = new DataOutputStream(fdos);
    for (int i = 0; i < 10; i++) {
        dos.writeLong(ThreadLocalRandom.current().nextLong());
    }
    try {
        SavepointStore.loadSavepoint(filePath.toString(), Thread.currentThread().getContextClassLoader());
        fail("Did not throw expected Exception");
    } catch (RuntimeException e) {
        assertTrue(e.getMessage().contains("Flink 1.0") && e.getMessage().contains("Unexpected magic number"));
    }
}

Example 17 with FSDataOutputStream

use of org.apache.flink.core.fs.FSDataOutputStream in project flink by apache.

the class DistCp method main.

public static void main(String[] args) throws Exception {
    // set up the execution environment
    final ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();
    ParameterTool params = ParameterTool.fromArgs(args);
    if (!params.has("input") || !params.has("output")) {
        System.err.println("Usage: --input <path> --output <path> [--parallelism <n>]");
        return;
    }
    final Path sourcePath = new Path(params.get("input"));
    final Path targetPath = new Path(params.get("output"));
    if (!isLocal(env) && !(isOnDistributedFS(sourcePath) && isOnDistributedFS(targetPath))) {
        System.out.println("In a distributed mode only HDFS input/output paths are supported");
        return;
    }
    final int parallelism = params.getInt("parallelism", 10);
    if (parallelism <= 0) {
        System.err.println("Parallelism should be greater than 0");
        return;
    }
    // make parameters available in the web interface
    env.getConfig().setGlobalJobParameters(params);
    env.setParallelism(parallelism);
    long startTime = System.currentTimeMillis();
    LOGGER.info("Initializing copy tasks");
    List<FileCopyTask> tasks = getCopyTasks(sourcePath);
    LOGGER.info("Copy task initialization took " + (System.currentTimeMillis() - startTime) + "ms");
    DataSet<FileCopyTask> inputTasks = new DataSource<>(env, new FileCopyTaskInputFormat(tasks), new GenericTypeInfo<>(FileCopyTask.class), "fileCopyTasks");
    FlatMapOperator<FileCopyTask, Object> res = inputTasks.flatMap(new RichFlatMapFunction<FileCopyTask, Object>() {

        private static final long serialVersionUID = 1109254230243989929L;

        private LongCounter fileCounter;

        private LongCounter bytesCounter;

        @Override
        public void open(Configuration parameters) throws Exception {
            bytesCounter = getRuntimeContext().getLongCounter(BYTES_COPIED_CNT_NAME);
            fileCounter = getRuntimeContext().getLongCounter(FILES_COPIED_CNT_NAME);
        }

        @Override
        public void flatMap(FileCopyTask task, Collector<Object> out) throws Exception {
            LOGGER.info("Processing task: " + task);
            Path outPath = new Path(targetPath, task.getRelativePath());
            FileSystem targetFs = targetPath.getFileSystem();
            // creating parent folders in case of a local FS
            if (!targetFs.isDistributedFS()) {
                // dealing with cases like file:///tmp or just /tmp
                File outFile = outPath.toUri().isAbsolute() ? new File(outPath.toUri()) : new File(outPath.toString());
                File parentFile = outFile.getParentFile();
                if (!parentFile.mkdirs() && !parentFile.exists()) {
                    throw new RuntimeException("Cannot create local file system directories: " + parentFile);
                }
            }
            FSDataOutputStream outputStream = null;
            FSDataInputStream inputStream = null;
            try {
                outputStream = targetFs.create(outPath, FileSystem.WriteMode.OVERWRITE);
                inputStream = task.getPath().getFileSystem().open(task.getPath());
                int bytes = IOUtils.copy(inputStream, outputStream);
                bytesCounter.add(bytes);
            } finally {
                IOUtils.closeQuietly(inputStream);
                IOUtils.closeQuietly(outputStream);
            }
            fileCounter.add(1L);
        }
    });
    // no data sinks are needed, therefore just printing an empty result
    res.print();
    Map<String, Object> accumulators = env.getLastJobExecutionResult().getAllAccumulatorResults();
    LOGGER.info("== COUNTERS ==");
    for (Map.Entry<String, Object> e : accumulators.entrySet()) {
        LOGGER.info(e.getKey() + ": " + e.getValue());
    }
}

Example 18 with FSDataOutputStream

use of org.apache.flink.core.fs.FSDataOutputStream in project flink by apache.

the class FileUtilsTest method generateTestFile.

/**
 * Generates a random content file.
 *
 * @param outputFile the path of the output file
 * @param length the size of content to generate
 * @return MD5 of the output file
 * @throws IOException
 * @throws NoSuchAlgorithmException
 */
private static String generateTestFile(String outputFile, int length) throws IOException, NoSuchAlgorithmException {
    Path outputFilePath = new Path(outputFile);
    final FileSystem fileSystem = outputFilePath.getFileSystem();
    try (final FSDataOutputStream fsDataOutputStream = fileSystem.create(outputFilePath, FileSystem.WriteMode.OVERWRITE)) {
        return writeRandomContent(fsDataOutputStream, length);
    }
}

Example 19 with FSDataOutputStream

use of org.apache.flink.core.fs.FSDataOutputStream in project flink by apache.

the class AbstractHadoopFileSystemITTest method testSimpleFileWriteAndRead.

@Test
public void testSimpleFileWriteAndRead() throws Exception {
    final String testLine = "Hello Upload!";
    final Path path = new Path(basePath, "test.txt");
    try {
        try (FSDataOutputStream out = fs.create(path, FileSystem.WriteMode.OVERWRITE);
            OutputStreamWriter writer = new OutputStreamWriter(out, StandardCharsets.UTF_8)) {
            writer.write(testLine);
        }
        // just in case, wait for the path to exist
        checkPathExistence(path, true, consistencyToleranceNS);
        try (FSDataInputStream in = fs.open(path);
            InputStreamReader ir = new InputStreamReader(in, StandardCharsets.UTF_8);
            BufferedReader reader = new BufferedReader(ir)) {
            String line = reader.readLine();
            assertEquals(testLine, line);
        }
    } finally {
        fs.delete(path, false);
    }
    checkPathExistence(path, false, consistencyToleranceNS);
}

Example 20 with FSDataOutputStream

use of org.apache.flink.core.fs.FSDataOutputStream in project flink by apache.

the class StateChangeFsUploader method wrap.

private OutputStreamWithPos wrap(FSDataOutputStream fsStream) throws IOException {
    StreamCompressionDecorator instance = compression ? SnappyStreamCompressionDecorator.INSTANCE : UncompressedStreamCompressionDecorator.INSTANCE;
    OutputStream compressed = compression ? instance.decorateWithCompression(fsStream) : fsStream;
    return new OutputStreamWithPos(new BufferedOutputStream(compressed, bufferSize));
}