Search in sources :

Example 6 with Transactional

use of co.cask.cdap.api.Transactional in project cdap by caskdata.

the class FileMetadataCleanerTest method testScanAndDeleteNewMetadata.

@Test
public void testScanAndDeleteNewMetadata() 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(), null);
    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);
    FileMetadataCleaner fileMetadataCleaner = new FileMetadataCleaner(datasetManager, transactional);
    try {
        long currentTime = System.currentTimeMillis();
        long eventTimestamp = currentTime - 100;
        LogPathIdentifier logPathIdentifier = new LogPathIdentifier("testNs2", "testApp", "testFlow");
        LocationFactory locationFactory = injector.getInstance(LocationFactory.class);
        List<String> expected = new ArrayList<>();
        for (int i = 0; i < 100; i++) {
            Location location = locationFactory.create("testFlowFile" + i);
            // values : event time is 100ms behind current timestamp
            fileMetaDataWriter.writeMetaData(logPathIdentifier, eventTimestamp + i, currentTime + i, location);
            expected.add(location.toURI().getPath());
        }
        long tillTime = currentTime + 50;
        List<FileMetadataCleaner.DeletedEntry> deletedEntries = fileMetadataCleaner.scanAndGetFilesToDelete(tillTime, TRANSACTION_TIMEOUT);
        // we should have deleted 51 rows, till time is inclusive
        Assert.assertEquals(51, deletedEntries.size());
        int count = 0;
        for (FileMetadataCleaner.DeletedEntry deletedEntry : deletedEntries) {
            Assert.assertEquals(expected.get(count), deletedEntry.getPath());
            count += 1;
        }
        // now add 10 entries for spark
        logPathIdentifier = new LogPathIdentifier("testNs2", "testApp", "testSpark");
        expected = new ArrayList<>();
        for (int i = 0; i < 10; i++) {
            Location location = locationFactory.create("testSparkFile" + i);
            // values : event time is 100ms behind current timestamp
            fileMetaDataWriter.writeMetaData(logPathIdentifier, eventTimestamp + i, currentTime + i, location);
            expected.add(location.toURI().getPath());
        }
        // lets keep the same till time - this should only delete the spark entries now
        deletedEntries = fileMetadataCleaner.scanAndGetFilesToDelete(tillTime, TRANSACTION_TIMEOUT);
        // we should have deleted 51 rows, till time is inclusive
        Assert.assertEquals(10, deletedEntries.size());
        count = 0;
        for (FileMetadataCleaner.DeletedEntry deletedEntry : deletedEntries) {
            Assert.assertEquals(expected.get(count), deletedEntry.getPath());
            count += 1;
        }
        // now add 10 entries in mr context in time range 60-70
        logPathIdentifier = new LogPathIdentifier("testNs2", "testApp", "testMr");
        expected = new ArrayList<>();
        // flow should come up at the beginning in the expected list
        for (int i = 51; i <= 70; i++) {
            expected.add(locationFactory.create("testFlowFile" + i).toURI().getPath());
        }
        for (int i = 0; i < 10; i++) {
            Location location = locationFactory.create("testMrFile" + i);
            // values : event time is 100ms behind current timestamp
            fileMetaDataWriter.writeMetaData(logPathIdentifier, eventTimestamp + i, currentTime + i, location);
            expected.add(location.toURI().getPath());
        }
        List<String> nextExpected = new ArrayList<>();
        logPathIdentifier = new LogPathIdentifier("testNs2", "testApp", "testCustomAction");
        for (int i = 90; i < 100; i++) {
            Location location = locationFactory.create("testActionFile" + i);
            // values : event time is 100ms behind current timestamp
            fileMetaDataWriter.writeMetaData(logPathIdentifier, eventTimestamp + i, currentTime + i, location);
            nextExpected.add(location.toURI().getPath());
        }
        tillTime = currentTime + 70;
        // lets delete till 70.
        deletedEntries = fileMetadataCleaner.scanAndGetFilesToDelete(tillTime, TRANSACTION_TIMEOUT);
        // we should have deleted 51-70 files of flow and 0-9 files of spark files in that order and 0 files of action.
        Assert.assertEquals(30, deletedEntries.size());
        count = 0;
        for (FileMetadataCleaner.DeletedEntry deletedEntry : deletedEntries) {
            Assert.assertEquals(expected.get(count), deletedEntry.getPath());
            count += 1;
        }
        // now delete till currentTime + 100, this should delete all remaining entries.
        // custom action should come first and then flow entries
        tillTime = currentTime + 100;
        // lets delete till 100.
        deletedEntries = fileMetadataCleaner.scanAndGetFilesToDelete(tillTime, TRANSACTION_TIMEOUT);
        // we should have deleted 90-99 of custom action(10) 71-99 (29) files of flow.
        for (int i = 71; i < 100; i++) {
            nextExpected.add(locationFactory.create("testFlowFile" + i).toURI().getPath());
        }
        Assert.assertEquals(39, deletedEntries.size());
        count = 0;
        for (FileMetadataCleaner.DeletedEntry deletedEntry : deletedEntries) {
            Assert.assertEquals(nextExpected.get(count), deletedEntry.getPath());
            count += 1;
        }
        // now lets do a delete with till time  = currentTime + 1000, this should return empty result
        tillTime = currentTime + 1000;
        deletedEntries = fileMetadataCleaner.scanAndGetFilesToDelete(tillTime, TRANSACTION_TIMEOUT);
        Assert.assertEquals(0, deletedEntries.size());
    } finally {
        // cleanup meta
        cleanupMetadata(transactional, datasetManager);
    }
}
Also used : FileMetaDataWriter(co.cask.cdap.logging.meta.FileMetaDataWriter) DefaultDatasetManager(co.cask.cdap.data2.datafabric.dataset.DefaultDatasetManager) DatasetManager(co.cask.cdap.api.dataset.DatasetManager) ArrayList(java.util.ArrayList) 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) LogPathIdentifier(co.cask.cdap.logging.appender.system.LogPathIdentifier) Transactional(co.cask.cdap.api.Transactional) Location(org.apache.twill.filesystem.Location) LogLocation(co.cask.cdap.logging.write.LogLocation) Test(org.junit.Test)

Example 7 with Transactional

use of co.cask.cdap.api.Transactional 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(), null);
    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 8 with Transactional

use of co.cask.cdap.api.Transactional in project cdap by caskdata.

the class LogCleanerTest method testLogCleanup.

@Test
public void testLogCleanup() 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(), null);
    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));
    FileMetadataCleaner fileMetadataCleaner = new FileMetadataCleaner(datasetManager, transactional);
    LocationFactory locationFactory = injector.getInstance(LocationFactory.class);
    long currentTime = System.currentTimeMillis();
    LogPathIdentifier logPathIdentifier = new LogPathIdentifier("testNs", "testApp", "testEntity");
    FileMetaDataWriter fileMetaDataWriter = new FileMetaDataWriter(datasetManager, transactional);
    long startTime = currentTime - 5000;
    Location dirLocation = locationFactory.create("logs");
    dirLocation.mkdirs();
    // create 20 files, add them in past time range
    for (int i = 0; i < 20; i++) {
        Location location = dirLocation.append("test" + i);
        location.createNew();
        fileMetaDataWriter.writeMetaData(logPathIdentifier, startTime + i, startTime + i, location);
    }
    Assert.assertEquals(20, dirLocation.list().size());
    LogCleaner logCleaner = new LogCleaner(fileMetadataCleaner, locationFactory, 100, 60);
    logCleaner.run();
    FileMetaDataReader fileMetaDataReader = injector.getInstance(FileMetaDataReader.class);
    // all meta data should be deleted
    Assert.assertEquals(0, fileMetaDataReader.listFiles(logPathIdentifier, 0, System.currentTimeMillis()).size());
// we are not asserting file existence as the delete could fail and we don't guarantee file deletion.
}
Also used : FileMetaDataWriter(co.cask.cdap.logging.meta.FileMetaDataWriter) 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) 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) Test(org.junit.Test)

Aggregations

Transactional (co.cask.cdap.api.Transactional)8 DatasetManager (co.cask.cdap.api.dataset.DatasetManager)7 SystemDatasetInstantiator (co.cask.cdap.data.dataset.SystemDatasetInstantiator)7 DefaultDatasetManager (co.cask.cdap.data2.datafabric.dataset.DefaultDatasetManager)7 DatasetFramework (co.cask.cdap.data2.dataset2.DatasetFramework)7 MultiThreadDatasetCache (co.cask.cdap.data2.dataset2.MultiThreadDatasetCache)7 FileMetaDataWriter (co.cask.cdap.logging.meta.FileMetaDataWriter)7 TransactionSystemClient (org.apache.tephra.TransactionSystemClient)7 LocationFactory (org.apache.twill.filesystem.LocationFactory)7 Test (org.junit.Test)7 LogPathIdentifier (co.cask.cdap.logging.appender.system.LogPathIdentifier)6 Location (org.apache.twill.filesystem.Location)6 FileMetaDataReader (co.cask.cdap.logging.meta.FileMetaDataReader)5 LogLocation (co.cask.cdap.logging.write.LogLocation)5 LoggingContext (co.cask.cdap.common.logging.LoggingContext)2 FileMetaDataManager (co.cask.cdap.logging.write.FileMetaDataManager)2 ArrayList (java.util.ArrayList)2 LoggingEvent (ch.qos.logback.classic.spi.LoggingEvent)1 TxRunnable (co.cask.cdap.api.TxRunnable)1 DatasetContext (co.cask.cdap.api.data.DatasetContext)1