Examples with FSWindowDataManager org.apache.apex.malhar.lib.wal.FSWindowDataManager used on opensource projects

Example 16 with FSWindowDataManager

use of org.apache.apex.malhar.lib.wal.FSWindowDataManager in project apex-malhar by apache.

the class AbstractFileInputOperatorTest method testIdempotencyWithMultipleEmitTuples.

@Test
public void testIdempotencyWithMultipleEmitTuples() throws Exception {
    FileContext.getLocalFSFileContext().delete(new Path(new File(testMeta.dir).getAbsolutePath()), true);
    List<String> allLines = Lists.newArrayList();
    for (int file = 0; file < 2; file++) {
        List<String> lines = Lists.newArrayList();
        for (int line = 0; line < 2; line++) {
            lines.add("f" + file + "l" + line);
        }
        allLines.addAll(lines);
        FileUtils.write(new File(testMeta.dir, "file" + file), StringUtils.join(lines, '\n'));
    }
    LineByLineFileInputOperator oper = new LineByLineFileInputOperator();
    FSWindowDataManager manager = new FSWindowDataManager();
    manager.setStatePath(testMeta.dir + "/recovery");
    oper.setWindowDataManager(manager);
    CollectorTestSink<String> queryResults = new CollectorTestSink<String>();
    TestUtils.setSink(oper.output, queryResults);
    oper.setDirectory(testMeta.dir);
    oper.getScanner().setFilePatternRegexp(".*file[\\d]");
    oper.setup(testMeta.context);
    oper.beginWindow(0);
    for (int i = 0; i < 3; i++) {
        oper.emitTuples();
    }
    oper.endWindow();
    oper.teardown();
    List<String> beforeRecovery = Lists.newArrayList(queryResults.collectedTuples);
    queryResults.clear();
    // idempotency  part
    oper.setup(testMeta.context);
    oper.beginWindow(0);
    oper.endWindow();
    Assert.assertEquals("number tuples", 4, queryResults.collectedTuples.size());
    Assert.assertEquals("lines", beforeRecovery, queryResults.collectedTuples);
    oper.teardown();
}

Example 17 with FSWindowDataManager

use of org.apache.apex.malhar.lib.wal.FSWindowDataManager in project apex-malhar by apache.

the class AbstractFileInputOperatorTest method testWindowDataManagerPartitioning.

@Test
public void testWindowDataManagerPartitioning() throws Exception {
    LineByLineFileInputOperator oper = new LineByLineFileInputOperator();
    oper.getScanner().setFilePatternRegexp(".*partition([\\d]*)");
    oper.setDirectory(new File(testMeta.dir).getAbsolutePath());
    oper.setWindowDataManager(new FSWindowDataManager());
    oper.operatorId = 7;
    Path path = new Path(new File(testMeta.dir).getAbsolutePath());
    FileContext.getLocalFSFileContext().delete(path, true);
    for (int file = 0; file < 4; file++) {
        FileUtils.write(new File(testMeta.dir, "partition00" + file), "");
    }
    List<Partition<AbstractFileInputOperator<String>>> partitions = Lists.newArrayList();
    partitions.add(new DefaultPartition<AbstractFileInputOperator<String>>(oper));
    Collection<Partition<AbstractFileInputOperator<String>>> newPartitions = oper.definePartitions(partitions, new PartitioningContextImpl(null, 2));
    Assert.assertEquals(2, newPartitions.size());
    Assert.assertEquals(1, oper.getCurrentPartitions());
    List<FSWindowDataManager> storageManagers = Lists.newLinkedList();
    for (Partition<AbstractFileInputOperator<String>> p : newPartitions) {
        storageManagers.add((FSWindowDataManager) p.getPartitionedInstance().getWindowDataManager());
    }
    Assert.assertEquals("count of storage managers", 2, storageManagers.size());
    int countOfDeleteManagers = 0;
    FSWindowDataManager deleteManager = null;
    for (FSWindowDataManager storageManager : storageManagers) {
        if (storageManager.getDeletedOperators() != null) {
            countOfDeleteManagers++;
            deleteManager = storageManager;
        }
    }
    Assert.assertEquals("count of delete managers", 1, countOfDeleteManagers);
    Assert.assertNotNull("deleted operators manager", deleteManager);
    Assert.assertEquals("deleted operators", Sets.newHashSet(7), deleteManager.getDeletedOperators());
}

Example 18 with FSWindowDataManager

use of org.apache.apex.malhar.lib.wal.FSWindowDataManager in project apex-malhar by apache.

the class AbstractFileInputOperatorTest method testStateWithIdempotency.

@Test
public void testStateWithIdempotency() throws Exception {
    FileContext.getLocalFSFileContext().delete(new Path(new File(testMeta.dir).getAbsolutePath()), true);
    HashSet<String> allLines = Sets.newHashSet();
    for (int file = 0; file < 3; file++) {
        HashSet<String> lines = Sets.newHashSet();
        for (int line = 0; line < 2; line++) {
            lines.add("f" + file + "l" + line);
        }
        allLines.addAll(lines);
        FileUtils.write(new File(testMeta.dir, "file" + file), StringUtils.join(lines, '\n'));
    }
    LineByLineFileInputOperator oper = new LineByLineFileInputOperator();
    FSWindowDataManager manager = new FSWindowDataManager();
    manager.setStatePath(testMeta.dir + "/recovery");
    oper.setWindowDataManager(manager);
    CollectorTestSink<String> queryResults = new CollectorTestSink<String>();
    @SuppressWarnings({ "unchecked", "rawtypes" }) CollectorTestSink<Object> sink = (CollectorTestSink) queryResults;
    oper.output.setSink(sink);
    oper.setDirectory(testMeta.dir);
    oper.getScanner().setFilePatternRegexp(".*file[\\d]");
    oper.setup(testMeta.context);
    for (long wid = 0; wid < 4; wid++) {
        oper.beginWindow(wid);
        oper.emitTuples();
        oper.endWindow();
    }
    oper.teardown();
    sink.clear();
    // idempotency  part
    oper.pendingFiles.add(new File(testMeta.dir, "file0").getAbsolutePath());
    oper.failedFiles.add(new AbstractFileInputOperator.FailedFile(new File(testMeta.dir, "file1").getAbsolutePath(), 0));
    oper.unfinishedFiles.add(new AbstractFileInputOperator.FailedFile(new File(testMeta.dir, "file2").getAbsolutePath(), 0));
    oper.setup(testMeta.context);
    for (long wid = 0; wid < 4; wid++) {
        oper.beginWindow(wid);
        oper.endWindow();
    }
    Assert.assertTrue("pending state", !oper.pendingFiles.contains("file0"));
    for (AbstractFileInputOperator.FailedFile failedFile : oper.failedFiles) {
        Assert.assertTrue("failed state", !failedFile.path.equals("file1"));
    }
    for (AbstractFileInputOperator.FailedFile unfinishedFile : oper.unfinishedFiles) {
        Assert.assertTrue("unfinished state", !unfinishedFile.path.equals("file2"));
    }
    oper.teardown();
}

Example 19 with FSWindowDataManager

use of org.apache.apex.malhar.lib.wal.FSWindowDataManager in project apex-malhar by apache.

the class AbstractKuduOutputOperator method setup.

@Override
public void setup(Context.OperatorContext context) {
    super.setup(context);
    windowDataManager = getWindowDataManager();
    if (windowDataManager == null) {
        windowDataManager = new FSWindowDataManager();
    }
    windowDataManager.setup(context);
}

Example 20 with FSWindowDataManager

use of org.apache.apex.malhar.lib.wal.FSWindowDataManager in project apex-malhar by apache.

the class FileSplitterInputTest method testIdempotency.

@Test
public void testIdempotency() throws InterruptedException {
    FSWindowDataManager fsIdempotentStorageManager = new FSWindowDataManager();
    testMeta.fileSplitterInput.setWindowDataManager(fsIdempotentStorageManager);
    testMeta.fileSplitterInput.setup(testMeta.context);
    // will emit window 1 from data directory
    validateFileMetadataInWindow1();
    testMeta.fileMetadataSink.clear();
    testMeta.blockMetadataSink.clear();
    testMeta.fileSplitterInput.teardown();
    testMeta.fileSplitterInput = KryoCloneUtils.cloneObject(testMeta.fileSplitterInput);
    testMeta.resetSinks();
    testMeta.fileSplitterInput.setup(testMeta.context);
    testMeta.fileSplitterInput.beginWindow(1);
    Assert.assertEquals("Blocks", 12, testMeta.blockMetadataSink.collectedTuples.size());
    for (Object blockMetadata : testMeta.blockMetadataSink.collectedTuples) {
        BlockMetadata.FileBlockMetadata metadata = (BlockMetadata.FileBlockMetadata) blockMetadata;
        Assert.assertTrue("path: " + metadata.getFilePath(), testMeta.filePaths.contains(metadata.getFilePath()));
    }
    testMeta.fileSplitterInput.teardown();
}