Search in sources :

Example 1 with FileMetaDataManager

use of co.cask.cdap.logging.write.FileMetaDataManager in project cdap by caskdata.

the class FileMetadataTest method testFileMetadataReadWriteAcrossFormats.

@Test
public void testFileMetadataReadWriteAcrossFormats() throws Exception {
    DatasetFramework datasetFramework = injector.getInstance(DatasetFramework.class);
    DatasetManager datasetManager = new DefaultDatasetManager(datasetFramework, NamespaceId.SYSTEM, co.cask.cdap.common.service.RetryStrategies.noRetry());
    Transactional transactional = Transactions.createTransactionalWithRetry(Transactions.createTransactional(new MultiThreadDatasetCache(new SystemDatasetInstantiator(datasetFramework), injector.getInstance(TransactionSystemClient.class), NamespaceId.SYSTEM, ImmutableMap.<String, String>of(), null, null)), RetryStrategies.retryOnConflict(20, 100));
    FileMetaDataManager fileMetaDataManager = injector.getInstance(FileMetaDataManager.class);
    FileMetaDataWriter fileMetaDataWriter = new FileMetaDataWriter(datasetManager, transactional);
    LogPathIdentifier logPathIdentifier = new LogPathIdentifier(NamespaceId.DEFAULT.getNamespace(), "testApp", "testFlow");
    LocationFactory locationFactory = injector.getInstance(LocationFactory.class);
    Location location = locationFactory.create(TMP_FOLDER.newFolder().getPath()).append("/logs");
    long currentTime = System.currentTimeMillis();
    LoggingContext loggingContext = LoggingContextHelper.getLoggingContext(NamespaceId.DEFAULT.getNamespace(), "testApp", "testFlow");
    // 10 files in old format
    for (int i = 1; i <= 10; i++) {
        // i is the event time
        fileMetaDataManager.writeMetaData(loggingContext, currentTime + i, location.append("testFile" + Integer.toString(i)));
    }
    long eventTime = currentTime + 20;
    long newCurrentTime = currentTime + 100;
    // 10 files in new format
    for (int i = 1; i <= 10; i++) {
        fileMetaDataWriter.writeMetaData(logPathIdentifier, eventTime + i, newCurrentTime + i, location.append("testFileNew" + Integer.toString(i)));
    }
    // reader test
    FileMetaDataReader fileMetadataReader = injector.getInstance(FileMetaDataReader.class);
    // scan only in old files time range
    List<LogLocation> locations = fileMetadataReader.listFiles(logPathIdentifier, currentTime + 2, currentTime + 6);
    // should include files from currentTime (1..6)
    Assert.assertEquals(6, locations.size());
    for (LogLocation logLocation : locations) {
        Assert.assertEquals(LogLocation.VERSION_0, logLocation.getFrameworkVersion());
    }
    // scan only in new files time range
    locations = fileMetadataReader.listFiles(logPathIdentifier, eventTime + 2, eventTime + 6);
    // should include files from currentTime (1..6)
    Assert.assertEquals(6, locations.size());
    for (LogLocation logLocation : locations) {
        Assert.assertEquals(LogLocation.VERSION_1, logLocation.getFrameworkVersion());
    }
    // scan time range across formats
    locations = fileMetadataReader.listFiles(logPathIdentifier, currentTime + 2, eventTime + 6);
    // should include files from old range (1..10) and new range (1..6)
    Assert.assertEquals(16, locations.size());
    for (int i = 0; i < locations.size(); i++) {
        if (i < 10) {
            Assert.assertEquals(LogLocation.VERSION_0, locations.get(i).getFrameworkVersion());
            Assert.assertEquals(location.append("testFile" + Integer.toString(i + 1)), locations.get(i).getLocation());
        } else {
            Assert.assertEquals(LogLocation.VERSION_1, locations.get(i).getFrameworkVersion());
            Assert.assertEquals(location.append("testFileNew" + Integer.toString(i - 9)), locations.get(i).getLocation());
        }
    }
}
Also used : FileMetaDataWriter(co.cask.cdap.logging.meta.FileMetaDataWriter) LoggingContext(co.cask.cdap.common.logging.LoggingContext) DefaultDatasetManager(co.cask.cdap.data2.datafabric.dataset.DefaultDatasetManager) DatasetManager(co.cask.cdap.api.dataset.DatasetManager) DefaultDatasetManager(co.cask.cdap.data2.datafabric.dataset.DefaultDatasetManager) LocationFactory(org.apache.twill.filesystem.LocationFactory) DatasetFramework(co.cask.cdap.data2.dataset2.DatasetFramework) TransactionSystemClient(org.apache.tephra.TransactionSystemClient) MultiThreadDatasetCache(co.cask.cdap.data2.dataset2.MultiThreadDatasetCache) SystemDatasetInstantiator(co.cask.cdap.data.dataset.SystemDatasetInstantiator) LogLocation(co.cask.cdap.logging.write.LogLocation) FileMetaDataManager(co.cask.cdap.logging.write.FileMetaDataManager) LogPathIdentifier(co.cask.cdap.logging.appender.system.LogPathIdentifier) FileMetaDataReader(co.cask.cdap.logging.meta.FileMetaDataReader) Transactional(co.cask.cdap.api.Transactional) Location(org.apache.twill.filesystem.Location) LogLocation(co.cask.cdap.logging.write.LogLocation) Test(org.junit.Test)

Example 2 with FileMetaDataManager

use of co.cask.cdap.logging.write.FileMetaDataManager in project cdap by caskdata.

the class FileMetadataCleanerTest method testScanAndDeleteOldMetadata.

@Test
public void testScanAndDeleteOldMetadata() throws Exception {
    // use file meta data manager to write meta data in old format
    // use file meta writer to write meta data in new format
    // scan for old files and make sure we only get the old meta data entries.
    DatasetFramework datasetFramework = injector.getInstance(DatasetFramework.class);
    DatasetManager datasetManager = new DefaultDatasetManager(datasetFramework, NamespaceId.SYSTEM, co.cask.cdap.common.service.RetryStrategies.noRetry());
    Transactional transactional = Transactions.createTransactionalWithRetry(Transactions.createTransactional(new MultiThreadDatasetCache(new SystemDatasetInstantiator(datasetFramework), injector.getInstance(TransactionSystemClient.class), NamespaceId.SYSTEM, ImmutableMap.<String, String>of(), null, null)), RetryStrategies.retryOnConflict(20, 100));
    FileMetaDataWriter fileMetaDataWriter = new FileMetaDataWriter(datasetManager, transactional);
    FileMetaDataManager fileMetaDataManager = injector.getInstance(FileMetaDataManager.class);
    LoggingContext flowContext = LoggingContextHelper.getLoggingContext("testNs", "testApp", "testFlow", ProgramType.FLOW);
    long eventTimestamp = System.currentTimeMillis();
    LocationFactory locationFactory = injector.getInstance(LocationFactory.class);
    Location testLocation = locationFactory.create("testFile");
    try {
        // write 50 entries in old format
        for (int i = 0; i < 50; i++) {
            fileMetaDataManager.writeMetaData(flowContext, eventTimestamp + i, testLocation);
        }
        LoggingContext wflowContext = LoggingContextHelper.getLoggingContext("testNs", "testApp", "testWflow", ProgramType.WORKFLOW);
        fileMetaDataManager.writeMetaData(wflowContext, eventTimestamp, testLocation);
        LoggingContext mrContext = LoggingContextHelper.getLoggingContext("testNs", "testApp", "testMR", ProgramType.MAPREDUCE);
        fileMetaDataManager.writeMetaData(mrContext, eventTimestamp, testLocation);
        LoggingContext sparkContext = LoggingContextHelper.getLoggingContext("testNs", "testApp", "testSpark", ProgramType.SPARK);
        fileMetaDataManager.writeMetaData(sparkContext, eventTimestamp, testLocation);
        // write 50 entries in new format
        long newEventTime = eventTimestamp + 1000;
        long currentTime = newEventTime + 1000;
        LogPathIdentifier logPathIdentifier = new LogPathIdentifier("testNs", "testApp", "testFlow");
        for (int i = 50; i < 100; i++) {
            fileMetaDataWriter.writeMetaData(logPathIdentifier, newEventTime + i, currentTime + i, testLocation);
        }
        FileMetaDataReader fileMetaDataReader = injector.getInstance(FileMetaDataReader.class);
        Assert.assertEquals(50, fileMetaDataReader.listFiles(LoggingContextHelper.getLogPathIdentifier(flowContext), eventTimestamp - 1, eventTimestamp + 100).size());
        Assert.assertEquals(1, fileMetaDataReader.listFiles(LoggingContextHelper.getLogPathIdentifier(wflowContext), eventTimestamp - 1, eventTimestamp + 100).size());
        Assert.assertEquals(1, fileMetaDataReader.listFiles(LoggingContextHelper.getLogPathIdentifier(mrContext), eventTimestamp - 1, eventTimestamp + 100).size());
        Assert.assertEquals(1, fileMetaDataReader.listFiles(LoggingContextHelper.getLogPathIdentifier(sparkContext), eventTimestamp - 1, eventTimestamp + 100).size());
        FileMetadataCleaner fileMetadataCleaner = new FileMetadataCleaner(datasetManager, transactional);
        fileMetadataCleaner.scanAndDeleteOldMetaData(TRANSACTION_TIMEOUT, CUTOFF_TIME_TRANSACTION);
        // deleted all old metadata
        Assert.assertEquals(0, fileMetaDataReader.listFiles(logPathIdentifier, eventTimestamp - 1, eventTimestamp + 100).size());
        Assert.assertEquals(0, fileMetaDataReader.listFiles(LoggingContextHelper.getLogPathIdentifier(wflowContext), eventTimestamp - 1, eventTimestamp + 100).size());
        Assert.assertEquals(0, fileMetaDataReader.listFiles(LoggingContextHelper.getLogPathIdentifier(mrContext), eventTimestamp - 1, eventTimestamp + 100).size());
        Assert.assertEquals(0, fileMetaDataReader.listFiles(LoggingContextHelper.getLogPathIdentifier(sparkContext), eventTimestamp - 1, eventTimestamp + 100).size());
    } finally {
        // cleanup meta
        cleanupMetadata(transactional, datasetManager);
    }
}
Also used : FileMetaDataWriter(co.cask.cdap.logging.meta.FileMetaDataWriter) LoggingContext(co.cask.cdap.common.logging.LoggingContext) DefaultDatasetManager(co.cask.cdap.data2.datafabric.dataset.DefaultDatasetManager) DatasetManager(co.cask.cdap.api.dataset.DatasetManager) DefaultDatasetManager(co.cask.cdap.data2.datafabric.dataset.DefaultDatasetManager) LocationFactory(org.apache.twill.filesystem.LocationFactory) DatasetFramework(co.cask.cdap.data2.dataset2.DatasetFramework) TransactionSystemClient(org.apache.tephra.TransactionSystemClient) MultiThreadDatasetCache(co.cask.cdap.data2.dataset2.MultiThreadDatasetCache) SystemDatasetInstantiator(co.cask.cdap.data.dataset.SystemDatasetInstantiator) FileMetaDataManager(co.cask.cdap.logging.write.FileMetaDataManager) LogPathIdentifier(co.cask.cdap.logging.appender.system.LogPathIdentifier) FileMetaDataReader(co.cask.cdap.logging.meta.FileMetaDataReader) Transactional(co.cask.cdap.api.Transactional) Location(org.apache.twill.filesystem.Location) LogLocation(co.cask.cdap.logging.write.LogLocation) Test(org.junit.Test)

Aggregations

Transactional (co.cask.cdap.api.Transactional)2 DatasetManager (co.cask.cdap.api.dataset.DatasetManager)2 LoggingContext (co.cask.cdap.common.logging.LoggingContext)2 SystemDatasetInstantiator (co.cask.cdap.data.dataset.SystemDatasetInstantiator)2 DefaultDatasetManager (co.cask.cdap.data2.datafabric.dataset.DefaultDatasetManager)2 DatasetFramework (co.cask.cdap.data2.dataset2.DatasetFramework)2 MultiThreadDatasetCache (co.cask.cdap.data2.dataset2.MultiThreadDatasetCache)2 LogPathIdentifier (co.cask.cdap.logging.appender.system.LogPathIdentifier)2 FileMetaDataReader (co.cask.cdap.logging.meta.FileMetaDataReader)2 FileMetaDataWriter (co.cask.cdap.logging.meta.FileMetaDataWriter)2 FileMetaDataManager (co.cask.cdap.logging.write.FileMetaDataManager)2 LogLocation (co.cask.cdap.logging.write.LogLocation)2 TransactionSystemClient (org.apache.tephra.TransactionSystemClient)2 Location (org.apache.twill.filesystem.Location)2 LocationFactory (org.apache.twill.filesystem.LocationFactory)2 Test (org.junit.Test)2