Search in sources :

Example 61 with LocalFileSystem

use of org.apache.hadoop.fs.LocalFileSystem in project hive by apache.

the class TestDruidStorageHandler method testCommitInsertIntoWhenDestinationSegmentFileExist.

@Test
public void testCommitInsertIntoWhenDestinationSegmentFileExist() throws MetaException, IOException {
    DerbyConnectorTestUtility connector = derbyConnectorRule.getConnector();
    MetadataStorageTablesConfig metadataStorageTablesConfig = derbyConnectorRule.metadataTablesConfigSupplier().get();
    druidStorageHandler.preCreateTable(tableMock);
    LocalFileSystem localFileSystem = FileSystem.getLocal(config);
    Path taskDirPath = new Path(tableWorkingPath, druidStorageHandler.makeStagingName());
    List<DataSegment> existingSegments = Collections.singletonList(createSegment(new Path(taskDirPath, "index_old.zip").toString(), new Interval(100, 150, DateTimeZone.UTC), "v0", new LinearShardSpec(1)));
    HdfsDataSegmentPusherConfig pusherConfig = new HdfsDataSegmentPusherConfig();
    pusherConfig.setStorageDirectory(config.get(String.valueOf(HiveConf.ConfVars.DRUID_SEGMENT_DIRECTORY)));
    DataSegmentPusher dataSegmentPusher = new HdfsDataSegmentPusher(pusherConfig, config, DruidStorageHandlerUtils.JSON_MAPPER);
    DruidStorageHandlerUtils.publishSegmentsAndCommit(connector, metadataStorageTablesConfig, DATA_SOURCE_NAME, existingSegments, true, config, dataSegmentPusher);
    DataSegment dataSegment = createSegment(new Path(taskDirPath, "index.zip").toString(), new Interval(100, 150, DateTimeZone.UTC), "v1", new LinearShardSpec(0));
    Path descriptorPath = DruidStorageHandlerUtils.makeSegmentDescriptorOutputPath(dataSegment, new Path(taskDirPath, DruidStorageHandler.SEGMENTS_DESCRIPTOR_DIR_NAME));
    DruidStorageHandlerUtils.writeSegmentDescriptor(localFileSystem, dataSegment, descriptorPath);
    // Create segment file at the destination location with LinearShardSpec(2)
    DataSegment segment = createSegment(new Path(taskDirPath, "index_conflict.zip").toString(), new Interval(100, 150, DateTimeZone.UTC), "v1", new LinearShardSpec(1));
    Path segmentPath = new Path(dataSegmentPusher.getPathForHadoop(), dataSegmentPusher.makeIndexPathName(segment, DruidStorageHandlerUtils.INDEX_ZIP));
    FileUtils.writeStringToFile(new File(segmentPath.toUri()), "dummy");
    druidStorageHandler.commitInsertTable(tableMock, false);
    Assert.assertArrayEquals(Lists.newArrayList(DATA_SOURCE_NAME).toArray(), Lists.newArrayList(DruidStorageHandlerUtils.getAllDataSourceNames(connector, metadataStorageTablesConfig)).toArray());
    final List<DataSegment> dataSegmentList = getUsedSegmentsList(connector, metadataStorageTablesConfig);
    Assert.assertEquals(2, dataSegmentList.size());
    DataSegment persistedSegment = dataSegmentList.get(1);
    // Insert into appends to old version
    Assert.assertEquals("v0", persistedSegment.getVersion());
    Assert.assertTrue(persistedSegment.getShardSpec() instanceof LinearShardSpec);
    // insert into should skip and increment partition number to 3
    Assert.assertEquals(2, persistedSegment.getShardSpec().getPartitionNum());
    Path expectedFinalHadoopPath = new Path(dataSegmentPusher.getPathForHadoop(), dataSegmentPusher.makeIndexPathName(persistedSegment, DruidStorageHandlerUtils.INDEX_ZIP));
    Assert.assertEquals(ImmutableMap.of("type", "hdfs", "path", expectedFinalHadoopPath.toString()), persistedSegment.getLoadSpec());
    Assert.assertEquals("dummySegmentData", FileUtils.readFileToString(new File(expectedFinalHadoopPath.toUri())));
}
Also used : Path(org.apache.hadoop.fs.Path) MetadataStorageTablesConfig(org.apache.druid.metadata.MetadataStorageTablesConfig) DataSegmentPusher(org.apache.druid.segment.loading.DataSegmentPusher) HdfsDataSegmentPusher(org.apache.druid.storage.hdfs.HdfsDataSegmentPusher) LocalFileSystem(org.apache.hadoop.fs.LocalFileSystem) LinearShardSpec(org.apache.druid.timeline.partition.LinearShardSpec) HdfsDataSegmentPusherConfig(org.apache.druid.storage.hdfs.HdfsDataSegmentPusherConfig) DataSegment(org.apache.druid.timeline.DataSegment) HdfsDataSegmentPusher(org.apache.druid.storage.hdfs.HdfsDataSegmentPusher) File(java.io.File) Interval(org.joda.time.Interval) Test(org.junit.Test)

Example 62 with LocalFileSystem

use of org.apache.hadoop.fs.LocalFileSystem in project hive by apache.

the class TestDruidStorageHandler method testCommitInsertIntoWithConflictingIntervalSegment.

@Test(expected = MetaException.class)
public void testCommitInsertIntoWithConflictingIntervalSegment() throws MetaException, IOException {
    DerbyConnectorTestUtility connector = derbyConnectorRule.getConnector();
    MetadataStorageTablesConfig metadataStorageTablesConfig = derbyConnectorRule.metadataTablesConfigSupplier().get();
    druidStorageHandler.preCreateTable(tableMock);
    LocalFileSystem localFileSystem = FileSystem.getLocal(config);
    Path taskDirPath = new Path(tableWorkingPath, druidStorageHandler.makeStagingName());
    List<DataSegment> existingSegments = Arrays.asList(createSegment(new Path(taskDirPath, "index_old_1.zip").toString(), new Interval(100, 150, DateTimeZone.UTC), "v0", new LinearShardSpec(0)), createSegment(new Path(taskDirPath, "index_old_2.zip").toString(), new Interval(150, 200, DateTimeZone.UTC), "v0", new LinearShardSpec(0)), createSegment(new Path(taskDirPath, "index_old_3.zip").toString(), new Interval(200, 300, DateTimeZone.UTC), "v0", new LinearShardSpec(0)));
    HdfsDataSegmentPusherConfig pusherConfig = new HdfsDataSegmentPusherConfig();
    pusherConfig.setStorageDirectory(taskDirPath.toString());
    DataSegmentPusher dataSegmentPusher = new HdfsDataSegmentPusher(pusherConfig, config, DruidStorageHandlerUtils.JSON_MAPPER);
    DruidStorageHandlerUtils.publishSegmentsAndCommit(connector, metadataStorageTablesConfig, DATA_SOURCE_NAME, existingSegments, true, config, dataSegmentPusher);
    // Try appending segment with conflicting interval
    DataSegment conflictingSegment = createSegment(new Path(taskDirPath, DruidStorageHandlerUtils.INDEX_ZIP).toString(), new Interval(100, 300, DateTimeZone.UTC), "v1", new LinearShardSpec(0));
    Path descriptorPath = DruidStorageHandlerUtils.makeSegmentDescriptorOutputPath(conflictingSegment, new Path(taskDirPath, DruidStorageHandler.SEGMENTS_DESCRIPTOR_DIR_NAME));
    DruidStorageHandlerUtils.writeSegmentDescriptor(localFileSystem, conflictingSegment, descriptorPath);
    druidStorageHandler.commitInsertTable(tableMock, false);
}
Also used : Path(org.apache.hadoop.fs.Path) MetadataStorageTablesConfig(org.apache.druid.metadata.MetadataStorageTablesConfig) DataSegmentPusher(org.apache.druid.segment.loading.DataSegmentPusher) HdfsDataSegmentPusher(org.apache.druid.storage.hdfs.HdfsDataSegmentPusher) LocalFileSystem(org.apache.hadoop.fs.LocalFileSystem) LinearShardSpec(org.apache.druid.timeline.partition.LinearShardSpec) HdfsDataSegmentPusherConfig(org.apache.druid.storage.hdfs.HdfsDataSegmentPusherConfig) DataSegment(org.apache.druid.timeline.DataSegment) HdfsDataSegmentPusher(org.apache.druid.storage.hdfs.HdfsDataSegmentPusher) Interval(org.joda.time.Interval) Test(org.junit.Test)

Example 63 with LocalFileSystem

use of org.apache.hadoop.fs.LocalFileSystem in project hive by apache.

the class TestDruidStorageHandler method testCommitInsertTable.

@Test
public void testCommitInsertTable() throws MetaException, IOException {
    druidStorageHandler.preCreateTable(tableMock);
    LocalFileSystem localFileSystem = FileSystem.getLocal(config);
    Path taskDirPath = new Path(tableWorkingPath, druidStorageHandler.makeStagingName());
    DataSegment dataSegment = createSegment(new Path(taskDirPath, "index.zip").toString());
    Path descriptorPath = DruidStorageHandlerUtils.makeSegmentDescriptorOutputPath(dataSegment, new Path(taskDirPath, DruidStorageHandler.SEGMENTS_DESCRIPTOR_DIR_NAME));
    DruidStorageHandlerUtils.writeSegmentDescriptor(localFileSystem, dataSegment, descriptorPath);
    druidStorageHandler.commitCreateTable(tableMock);
    Assert.assertArrayEquals(Lists.newArrayList(DATA_SOURCE_NAME).toArray(), Lists.newArrayList(DruidStorageHandlerUtils.getAllDataSourceNames(derbyConnectorRule.getConnector(), derbyConnectorRule.metadataTablesConfigSupplier().get())).toArray());
}
Also used : Path(org.apache.hadoop.fs.Path) LocalFileSystem(org.apache.hadoop.fs.LocalFileSystem) DataSegment(org.apache.druid.timeline.DataSegment) Test(org.junit.Test)

Example 64 with LocalFileSystem

use of org.apache.hadoop.fs.LocalFileSystem in project hive by apache.

the class TestDruidStorageHandler method testCommitInsertOverwriteTable.

@Test
public void testCommitInsertOverwriteTable() throws MetaException, IOException {
    DerbyConnectorTestUtility connector = derbyConnectorRule.getConnector();
    MetadataStorageTablesConfig metadataStorageTablesConfig = derbyConnectorRule.metadataTablesConfigSupplier().get();
    druidStorageHandler.preCreateTable(tableMock);
    LocalFileSystem localFileSystem = FileSystem.getLocal(config);
    Path taskDirPath = new Path(tableWorkingPath, druidStorageHandler.makeStagingName());
    HdfsDataSegmentPusherConfig pusherConfig = new HdfsDataSegmentPusherConfig();
    pusherConfig.setStorageDirectory(config.get(String.valueOf(HiveConf.ConfVars.DRUID_SEGMENT_DIRECTORY)));
    DataSegmentPusher dataSegmentPusher = new HdfsDataSegmentPusher(pusherConfig, config, DruidStorageHandlerUtils.JSON_MAPPER);
    // This create and publish the segment to be overwritten
    List<DataSegment> existingSegments = Collections.singletonList(createSegment(new Path(taskDirPath, DruidStorageHandlerUtils.INDEX_ZIP).toString(), new Interval(100, 150, DateTimeZone.UTC), "v0", new LinearShardSpec(0)));
    DruidStorageHandlerUtils.publishSegmentsAndCommit(connector, metadataStorageTablesConfig, DATA_SOURCE_NAME, existingSegments, true, config, dataSegmentPusher);
    // This creates and publish new segment
    DataSegment dataSegment = createSegment(new Path(taskDirPath, DruidStorageHandlerUtils.INDEX_ZIP).toString(), new Interval(180, 250, DateTimeZone.UTC), "v1", new LinearShardSpec(0));
    Path descriptorPath = DruidStorageHandlerUtils.makeSegmentDescriptorOutputPath(dataSegment, new Path(taskDirPath, DruidStorageHandler.SEGMENTS_DESCRIPTOR_DIR_NAME));
    DruidStorageHandlerUtils.writeSegmentDescriptor(localFileSystem, dataSegment, descriptorPath);
    druidStorageHandler.commitInsertTable(tableMock, true);
    Assert.assertArrayEquals(Lists.newArrayList(DATA_SOURCE_NAME).toArray(), Lists.newArrayList(DruidStorageHandlerUtils.getAllDataSourceNames(connector, metadataStorageTablesConfig)).toArray());
    final List<DataSegment> dataSegmentList = getUsedSegmentsList(connector, metadataStorageTablesConfig);
    Assert.assertEquals(1, dataSegmentList.size());
    DataSegment persistedSegment = Iterables.getOnlyElement(dataSegmentList);
    Assert.assertEquals(dataSegment, persistedSegment);
    Assert.assertEquals(dataSegment.getVersion(), persistedSegment.getVersion());
    Path expectedFinalHadoopPath = new Path(dataSegmentPusher.getPathForHadoop(), dataSegmentPusher.makeIndexPathName(persistedSegment, DruidStorageHandlerUtils.INDEX_ZIP));
    Assert.assertEquals(ImmutableMap.of("type", "hdfs", "path", expectedFinalHadoopPath.toString()), persistedSegment.getLoadSpec());
    Assert.assertEquals("dummySegmentData", FileUtils.readFileToString(new File(expectedFinalHadoopPath.toUri())));
}
Also used : Path(org.apache.hadoop.fs.Path) MetadataStorageTablesConfig(org.apache.druid.metadata.MetadataStorageTablesConfig) DataSegmentPusher(org.apache.druid.segment.loading.DataSegmentPusher) HdfsDataSegmentPusher(org.apache.druid.storage.hdfs.HdfsDataSegmentPusher) LocalFileSystem(org.apache.hadoop.fs.LocalFileSystem) LinearShardSpec(org.apache.druid.timeline.partition.LinearShardSpec) HdfsDataSegmentPusherConfig(org.apache.druid.storage.hdfs.HdfsDataSegmentPusherConfig) HdfsDataSegmentPusher(org.apache.druid.storage.hdfs.HdfsDataSegmentPusher) DataSegment(org.apache.druid.timeline.DataSegment) File(java.io.File) Interval(org.joda.time.Interval) Test(org.junit.Test)

Example 65 with LocalFileSystem

use of org.apache.hadoop.fs.LocalFileSystem in project hive by apache.

the class TestDruidStorageHandler method testCommitCreateTablePlusCommitDropTableWithPurge.

@Test
public void testCommitCreateTablePlusCommitDropTableWithPurge() throws MetaException, IOException {
    druidStorageHandler.preCreateTable(tableMock);
    LocalFileSystem localFileSystem = FileSystem.getLocal(config);
    Path taskDirPath = new Path(tableWorkingPath, druidStorageHandler.makeStagingName());
    DataSegment dataSegment = createSegment(new Path(taskDirPath, "index.zip").toString());
    Path descriptorPath = DruidStorageHandlerUtils.makeSegmentDescriptorOutputPath(dataSegment, new Path(taskDirPath, DruidStorageHandler.SEGMENTS_DESCRIPTOR_DIR_NAME));
    DruidStorageHandlerUtils.writeSegmentDescriptor(localFileSystem, dataSegment, descriptorPath);
    druidStorageHandler.commitCreateTable(tableMock);
    Assert.assertArrayEquals(Lists.newArrayList(DATA_SOURCE_NAME).toArray(), Lists.newArrayList(DruidStorageHandlerUtils.getAllDataSourceNames(derbyConnectorRule.getConnector(), derbyConnectorRule.metadataTablesConfigSupplier().get())).toArray());
    druidStorageHandler.commitDropTable(tableMock, true);
    Assert.assertArrayEquals(Lists.newArrayList().toArray(), Lists.newArrayList(DruidStorageHandlerUtils.getAllDataSourceNames(derbyConnectorRule.getConnector(), derbyConnectorRule.metadataTablesConfigSupplier().get())).toArray());
}
Also used : Path(org.apache.hadoop.fs.Path) LocalFileSystem(org.apache.hadoop.fs.LocalFileSystem) DataSegment(org.apache.druid.timeline.DataSegment) Test(org.junit.Test)

Aggregations

LocalFileSystem (org.apache.hadoop.fs.LocalFileSystem)120 Path (org.apache.hadoop.fs.Path)77 Test (org.junit.Test)63 Configuration (org.apache.hadoop.conf.Configuration)56 FileSystem (org.apache.hadoop.fs.FileSystem)35 IOException (java.io.IOException)33 File (java.io.File)23 NewTableConfiguration (org.apache.accumulo.core.client.admin.NewTableConfiguration)23 SamplerConfiguration (org.apache.accumulo.core.client.sample.SamplerConfiguration)23 SummarizerConfiguration (org.apache.accumulo.core.client.summary.SummarizerConfiguration)23 DefaultConfiguration (org.apache.accumulo.core.conf.DefaultConfiguration)23 Key (org.apache.accumulo.core.data.Key)22 Value (org.apache.accumulo.core.data.Value)22 ArrayList (java.util.ArrayList)19 ExecutorService (java.util.concurrent.ExecutorService)15 Future (java.util.concurrent.Future)15 Scanner (org.apache.accumulo.core.client.Scanner)14 DataSegment (org.apache.druid.timeline.DataSegment)13 DataSegmentPusher (org.apache.druid.segment.loading.DataSegmentPusher)8 HdfsDataSegmentPusher (org.apache.druid.storage.hdfs.HdfsDataSegmentPusher)8