Search in sources :

Example 56 with UserClientConnection

use of org.apache.drill.exec.rpc.UserClientConnection in project drill by axbaretto.

the class TestRecordIterator method testMarkResetIterator.

@Test
public void testMarkResetIterator() throws Throwable {
    final DrillbitContext bitContext = mockDrillbitContext();
    final UserClientConnection connection = Mockito.mock(UserClientConnection.class);
    final PhysicalPlanReader reader = PhysicalPlanReaderTestFactory.defaultPhysicalPlanReader(c);
    final String planStr = Files.toString(DrillFileUtils.getResourceAsFile("/record/test_recorditerator.json"), Charsets.UTF_8);
    final PhysicalPlan plan = reader.readPhysicalPlan(planStr);
    final FunctionImplementationRegistry registry = new FunctionImplementationRegistry(c);
    final FragmentContextImpl context = new FragmentContextImpl(bitContext, BitControl.PlanFragment.getDefaultInstance(), connection, registry);
    final List<PhysicalOperator> operatorList = plan.getSortedOperators(false);
    SimpleRootExec exec = new SimpleRootExec(ImplCreator.getExec(context, (FragmentRoot) operatorList.iterator().next()));
    RecordBatch singleBatch = exec.getIncoming();
    PhysicalOperator dummyPop = operatorList.iterator().next();
    OpProfileDef def = new OpProfileDef(dummyPop.getOperatorId(), UserBitShared.CoreOperatorType.MOCK_SUB_SCAN_VALUE, OperatorUtilities.getChildCount(dummyPop));
    OperatorStats stats = exec.getContext().getStats().newOperatorStats(def, exec.getContext().getAllocator());
    RecordIterator iter = new RecordIterator(singleBatch, null, exec.getContext().newOperatorContext(dummyPop, stats), 0, null);
    List<ValueVector> vectors = null;
    // batche sizes
    // 1, 100, 10, 10000, 1, 1000
    // total = 11112
    // BATCH 1 : 1, starting outerposition: 0
    iter.next();
    assertFalse(iter.finished());
    assertEquals(1, iter.getTotalRecordCount());
    assertEquals(0, iter.getCurrentPosition());
    assertEquals(0, iter.getOuterPosition());
    assertEquals(1, iter.cachedBatches().size());
    vectors = Lists.newArrayList();
    for (VectorWrapper vw : iter) {
        vectors.add(vw.getValueVector());
    }
    // mark at position 0
    iter.mark();
    checkValues(vectors, 0);
    // BATCH 2: 100, starting outerposition: 1
    iter.next();
    assertFalse(iter.finished());
    assertEquals(101, iter.getTotalRecordCount(), 101);
    assertEquals(0, iter.getCurrentPosition());
    assertEquals(100, iter.getInnerRecordCount());
    assertEquals(1, iter.getOuterPosition());
    assertEquals(2, iter.cachedBatches().size());
    for (int i = 0; i < 100; i++) {
        checkValues(vectors, i);
        iter.next();
    }
    // BATCH 3 :10, starting outerposition: 101
    assertFalse(iter.finished());
    assertEquals(111, iter.getTotalRecordCount());
    assertEquals(0, iter.getCurrentPosition());
    assertEquals(10, iter.getInnerRecordCount());
    assertEquals(101, iter.getOuterPosition());
    assertEquals(3, iter.cachedBatches().size());
    for (int i = 0; i < 10; i++) {
        checkValues(vectors, i);
        iter.next();
    }
    // BATCH 4 : 10000, starting outerposition: 111
    assertFalse(iter.finished());
    assertEquals(10111, iter.getTotalRecordCount());
    assertEquals(0, iter.getCurrentPosition(), 0);
    assertEquals(10000, iter.getInnerRecordCount());
    assertEquals(111, iter.getOuterPosition());
    assertEquals(4, iter.cachedBatches().size());
    for (int i = 0; i < 10000; i++) {
        checkValues(vectors, i);
        iter.next();
    }
    // BATCH 5 : 1, starting outerposition: 10111
    assertFalse(iter.finished());
    assertEquals(10112, iter.getTotalRecordCount());
    assertEquals(0, iter.getCurrentPosition());
    assertEquals(1, iter.getInnerRecordCount());
    assertEquals(10111, iter.getOuterPosition());
    assertEquals(5, iter.cachedBatches().size());
    checkValues(vectors, 0);
    iter.next();
    // BATCH 6 : 1000, starting outerposition: 10112
    assertFalse(iter.finished());
    assertEquals(11112, iter.getTotalRecordCount());
    assertEquals(0, iter.getCurrentPosition());
    assertEquals(1000, iter.getInnerRecordCount());
    assertEquals(10112, iter.getOuterPosition());
    assertEquals(6, iter.cachedBatches().size());
    for (int i = 0; i < 1000; i++) {
        checkValues(vectors, i);
        iter.next();
    }
    assertTrue(iter.finished());
    assertEquals(6, iter.cachedBatches().size());
    // back to batch 1
    iter.reset();
    assertFalse(iter.finished());
    assertEquals(iter.getTotalRecordCount(), 11112);
    assertEquals(6, iter.cachedBatches().size());
    assertEquals(iter.getCurrentPosition(), 0);
    assertEquals(1, iter.getInnerRecordCount());
    checkValues(vectors, 0);
    iter.next();
    // mark start of batch 2
    iter.mark();
    assertFalse(iter.finished());
    assertEquals(iter.getTotalRecordCount(), 11112);
    assertEquals(5, iter.cachedBatches().size());
    assertEquals(iter.getCurrentPosition(), 0);
    assertEquals(100, iter.getInnerRecordCount());
    for (int i = 0; i < 100; i++) {
        iter.next();
    }
    // mark start of batch 3
    iter.mark();
    assertFalse(iter.finished());
    assertEquals(iter.getTotalRecordCount(), 11112);
    assertEquals(4, iter.cachedBatches().size());
    assertEquals(iter.getCurrentPosition(), 0);
    assertEquals(10, iter.getInnerRecordCount());
    for (int i = 0; i < 10; i++) {
        iter.next();
    }
    // jump into middle of largest batch #4.
    for (int i = 0; i < 5000; i++) {
        iter.next();
    }
    assertEquals(4, iter.cachedBatches().size());
    iter.mark();
    assertEquals(3, iter.cachedBatches().size());
    for (int i = 0; i < 5000; i++) {
        iter.next();
    }
    // mark start of batch 5
    iter.mark();
    assertFalse(iter.finished());
    assertEquals(iter.getTotalRecordCount(), 11112);
    assertEquals(2, iter.cachedBatches().size());
    assertEquals(iter.getCurrentPosition(), 0);
    assertEquals(1, iter.getInnerRecordCount());
    // move to last batch
    iter.next();
    // skip to the middle of last batch
    for (int i = 0; i < 500; i++) {
        iter.next();
    }
    checkValues(vectors, 499);
    checkValues(vectors, 500);
    iter.reset();
    checkValues(vectors, 0);
    assertFalse(iter.finished());
    assertEquals(iter.getTotalRecordCount(), 11112);
    assertEquals(2, iter.cachedBatches().size());
    assertEquals(iter.getCurrentPosition(), 0);
    assertEquals(1, iter.getInnerRecordCount());
    // move to last batch
    iter.next();
    assertEquals(0, iter.getCurrentPosition());
    for (int i = 0; i < 500; i++) {
        iter.next();
    }
    // This should free 5th batch.
    iter.mark();
    assertFalse(iter.finished());
    assertEquals(iter.getTotalRecordCount(), 11112);
    assertEquals(1, iter.cachedBatches().size());
    assertEquals(500, iter.getCurrentPosition());
    assertEquals(1000, iter.getInnerRecordCount());
    // go to the end of iterator
    for (int i = 0; i < 500; i++) {
        iter.next();
    }
    assertTrue(iter.finished());
    iter.reset();
    assertFalse(iter.finished());
    assertEquals(iter.getTotalRecordCount(), 11112);
    assertEquals(1, iter.cachedBatches().size());
    assertEquals(500, iter.getCurrentPosition());
    assertEquals(1000, iter.getInnerRecordCount());
    iter.close();
    assertEquals(0, iter.cachedBatches().size());
}
Also used : DrillbitContext(org.apache.drill.exec.server.DrillbitContext) PhysicalPlan(org.apache.drill.exec.physical.PhysicalPlan) OpProfileDef(org.apache.drill.exec.ops.OpProfileDef) PhysicalPlanReader(org.apache.drill.exec.planner.PhysicalPlanReader) FragmentContextImpl(org.apache.drill.exec.ops.FragmentContextImpl) FragmentRoot(org.apache.drill.exec.physical.base.FragmentRoot) OperatorStats(org.apache.drill.exec.ops.OperatorStats) ValueVector(org.apache.drill.exec.vector.ValueVector) SimpleRootExec(org.apache.drill.exec.physical.impl.SimpleRootExec) PhysicalOperator(org.apache.drill.exec.physical.base.PhysicalOperator) UserClientConnection(org.apache.drill.exec.rpc.UserClientConnection) FunctionImplementationRegistry(org.apache.drill.exec.expr.fn.FunctionImplementationRegistry) Test(org.junit.Test) VectorTest(org.apache.drill.categories.VectorTest)

Example 57 with UserClientConnection

use of org.apache.drill.exec.rpc.UserClientConnection in project drill by apache.

the class ParquetRecordReaderTest method testPerformance.

@Test
@Ignore
public void testPerformance() throws Exception {
    final DrillbitContext bitContext = mock(DrillbitContext.class);
    final UserClientConnection connection = mock(UserClientConnection.class);
    final DrillConfig c = DrillConfig.create();
    final FunctionImplementationRegistry registry = new FunctionImplementationRegistry(c);
    final FragmentContextImpl context = new FragmentContextImpl(bitContext, BitControl.PlanFragment.getDefaultInstance(), connection, registry);
    final Path fileName = new Path("/tmp/parquet_test_performance.parquet");
    final HashMap<String, FieldInfo> fields = new HashMap<>();
    final ParquetTestProperties props = new ParquetTestProperties(1, 20 * 1000 * 1000, DEFAULT_BYTES_PER_PAGE, fields);
    populateFieldInfoMap(props);
    final Configuration dfsConfig = new Configuration();
    final List<Footer> footers = ParquetFileReader.readFooters(dfsConfig, fileName);
    final Footer f = footers.iterator().next();
    final List<SchemaPath> columns = Lists.newArrayList();
    columns.add(new SchemaPath("_MAP.integer", ExpressionPosition.UNKNOWN));
    columns.add(new SchemaPath("_MAP.bigInt", ExpressionPosition.UNKNOWN));
    columns.add(new SchemaPath("_MAP.f", ExpressionPosition.UNKNOWN));
    columns.add(new SchemaPath("_MAP.d", ExpressionPosition.UNKNOWN));
    columns.add(new SchemaPath("_MAP.b", ExpressionPosition.UNKNOWN));
    columns.add(new SchemaPath("_MAP.bin", ExpressionPosition.UNKNOWN));
    columns.add(new SchemaPath("_MAP.bin2", ExpressionPosition.UNKNOWN));
    int totalRowCount = 0;
    final FileSystem fs = new CachedSingleFileSystem(fileName);
    final BufferAllocator allocator = RootAllocatorFactory.newRoot(c);
    for (int i = 0; i < 25; i++) {
        CompressionCodecFactory ccf = DrillCompressionCodecFactory.createDirectCodecFactory(dfsConfig, new ParquetDirectByteBufferAllocator(allocator), 0);
        final ParquetRecordReader rr = new ParquetRecordReader(context, fileName, 0, fs, ccf, f.getParquetMetadata(), columns, ParquetReaderUtility.DateCorruptionStatus.META_SHOWS_CORRUPTION);
        final TestOutputMutator mutator = new TestOutputMutator(allocator);
        rr.setup(null, mutator);
        final Stopwatch watch = Stopwatch.createStarted();
        int rowCount = 0;
        while ((rowCount = rr.next()) > 0) {
            totalRowCount += rowCount;
        }
        rr.close();
    }
    allocator.close();
}
Also used : DrillbitContext(org.apache.drill.exec.server.DrillbitContext) Configuration(org.apache.hadoop.conf.Configuration) HashMap(java.util.HashMap) Stopwatch(org.apache.drill.shaded.guava.com.google.common.base.Stopwatch) FragmentContextImpl(org.apache.drill.exec.ops.FragmentContextImpl) TestOutputMutator(org.apache.drill.exec.store.TestOutputMutator) DrillConfig(org.apache.drill.common.config.DrillConfig) CompressionCodecFactory(org.apache.parquet.compression.CompressionCodecFactory) DrillCompressionCodecFactory(org.apache.drill.exec.store.parquet.compression.DrillCompressionCodecFactory) SchemaPath(org.apache.drill.common.expression.SchemaPath) FileSystem(org.apache.hadoop.fs.FileSystem) CachedSingleFileSystem(org.apache.drill.exec.store.CachedSingleFileSystem) FunctionImplementationRegistry(org.apache.drill.exec.expr.fn.FunctionImplementationRegistry) Path(org.apache.hadoop.fs.Path) SchemaPath(org.apache.drill.common.expression.SchemaPath) BufferAllocator(org.apache.drill.exec.memory.BufferAllocator) ParquetRecordReader(org.apache.drill.exec.store.parquet.columnreaders.ParquetRecordReader) CachedSingleFileSystem(org.apache.drill.exec.store.CachedSingleFileSystem) UserClientConnection(org.apache.drill.exec.rpc.UserClientConnection) Footer(org.apache.parquet.hadoop.Footer) Ignore(org.junit.Ignore) Test(org.junit.Test)

Example 58 with UserClientConnection

use of org.apache.drill.exec.rpc.UserClientConnection in project drill by apache.

the class DumpCatTest method testDumpCat.

@Test
public void testDumpCat() throws Throwable {
    final DrillbitContext bitContext = mockDrillbitContext();
    final UserClientConnection connection = Mockito.mock(UserClientConnection.class);
    final PhysicalPlanReader reader = defaultPhysicalPlanReader(c);
    final PhysicalPlan plan = reader.readPhysicalPlan(Files.asCharSource(DrillFileUtils.getResourceAsFile("/trace/simple_trace.json"), Charsets.UTF_8).read());
    final FunctionImplementationRegistry registry = new FunctionImplementationRegistry(c);
    final FragmentContextImpl context = new FragmentContextImpl(bitContext, PlanFragment.getDefaultInstance(), connection, registry);
    final SimpleRootExec exec = new SimpleRootExec(ImplCreator.getExec(context, (FragmentRoot) plan.getSortedOperators(false).iterator().next()));
    while (exec.next()) {
    }
    if (context.getExecutorState().getFailureCause() != null) {
        throw context.getExecutorState().getFailureCause();
    }
    assertTrue(!context.getExecutorState().isFailed());
    exec.close();
    FragmentHandle handle = context.getHandle();
    /* Form the file name to which the trace output will dump the record batches */
    String qid = QueryIdHelper.getQueryId(handle.getQueryId());
    final int majorFragmentId = handle.getMajorFragmentId();
    final int minorFragmentId = handle.getMinorFragmentId();
    final String logLocation = c.getString(ExecConstants.TRACE_DUMP_DIRECTORY);
    final String filename = String.format("%s//%s_%d_%d_mock-scan", logLocation, qid, majorFragmentId, minorFragmentId);
    final Configuration conf = new Configuration();
    conf.set(FileSystem.FS_DEFAULT_NAME_KEY, c.getString(ExecConstants.TRACE_DUMP_FILESYSTEM));
    final FileSystem fs = FileSystem.get(conf);
    final Path path = new Path(filename);
    assertTrue("Trace file does not exist", fs.exists(path));
    final DumpCat dumpCat = new DumpCat();
    // Test Query mode
    try (final FileInputStream input = new FileInputStream(filename)) {
        dumpCat.doQuery(input);
    }
    // Test Batch mode
    try (final FileInputStream input = new FileInputStream(filename)) {
        dumpCat.doBatch(input, 0, true);
    }
}
Also used : DrillbitContext(org.apache.drill.exec.server.DrillbitContext) Path(org.apache.hadoop.fs.Path) PhysicalPlan(org.apache.drill.exec.physical.PhysicalPlan) Configuration(org.apache.hadoop.conf.Configuration) PhysicalPlanReaderTestFactory.defaultPhysicalPlanReader(org.apache.drill.exec.planner.PhysicalPlanReaderTestFactory.defaultPhysicalPlanReader) PhysicalPlanReader(org.apache.drill.exec.planner.PhysicalPlanReader) FragmentContextImpl(org.apache.drill.exec.ops.FragmentContextImpl) FragmentRoot(org.apache.drill.exec.physical.base.FragmentRoot) FragmentHandle(org.apache.drill.exec.proto.ExecProtos.FragmentHandle) FileInputStream(java.io.FileInputStream) SimpleRootExec(org.apache.drill.exec.physical.impl.SimpleRootExec) FileSystem(org.apache.hadoop.fs.FileSystem) UserClientConnection(org.apache.drill.exec.rpc.UserClientConnection) FunctionImplementationRegistry(org.apache.drill.exec.expr.fn.FunctionImplementationRegistry) ExecTest(org.apache.drill.exec.ExecTest) Test(org.junit.Test)

Example 59 with UserClientConnection

use of org.apache.drill.exec.rpc.UserClientConnection in project drill by apache.

the class TestNewMathFunctions method runTest.

public void runTest(Object[] expectedResults, String planPath) throws Throwable {
    final DrillbitContext bitContext = mockDrillbitContext();
    final UserClientConnection connection = Mockito.mock(UserClientConnection.class);
    final String planString = Resources.toString(Resources.getResource(planPath), Charsets.UTF_8);
    if (reader == null) {
        reader = PhysicalPlanReaderTestFactory.defaultPhysicalPlanReader(c);
    }
    if (registry == null) {
        registry = new FunctionImplementationRegistry(c);
    }
    if (context == null) {
        context = new FragmentContextImpl(bitContext, PlanFragment.getDefaultInstance(), connection, registry);
    }
    final PhysicalPlan plan = reader.readPhysicalPlan(planString);
    final SimpleRootExec exec = new SimpleRootExec(ImplCreator.getExec(context, (FragmentRoot) plan.getSortedOperators(false).iterator().next()));
    while (exec.next()) {
        final Object[] res = getRunResult(exec);
        assertEquals("return count does not match", expectedResults.length, res.length);
        for (int i = 0; i < res.length; i++) {
            assertEquals(String.format("column %s does not match", i), res[i], expectedResults[i]);
        }
    }
    if (context.getExecutorState().getFailureCause() != null) {
        throw context.getExecutorState().getFailureCause();
    }
    assertTrue(!context.getExecutorState().isFailed());
}
Also used : DrillbitContext(org.apache.drill.exec.server.DrillbitContext) SimpleRootExec(org.apache.drill.exec.physical.impl.SimpleRootExec) PhysicalPlan(org.apache.drill.exec.physical.PhysicalPlan) UserClientConnection(org.apache.drill.exec.rpc.UserClientConnection) FragmentContextImpl(org.apache.drill.exec.ops.FragmentContextImpl) FragmentRoot(org.apache.drill.exec.physical.base.FragmentRoot) FunctionImplementationRegistry(org.apache.drill.exec.expr.fn.FunctionImplementationRegistry)

Example 60 with UserClientConnection

use of org.apache.drill.exec.rpc.UserClientConnection in project drill by apache.

the class TestRepeatedFunction method testRepeated.

@Test
public void testRepeated() throws Throwable {
    final DrillbitContext bitContext = mockDrillbitContext();
    final UserClientConnection connection = Mockito.mock(UserClientConnection.class);
    final PhysicalPlanReader reader = PhysicalPlanReaderTestFactory.defaultPhysicalPlanReader(c);
    final PhysicalPlan plan = reader.readPhysicalPlan(Files.asCharSource(DrillFileUtils.getResourceAsFile("/physical_repeated_1.json"), Charsets.UTF_8).read());
    final FunctionImplementationRegistry registry = new FunctionImplementationRegistry(c);
    final FragmentContextImpl context = new FragmentContextImpl(bitContext, PlanFragment.getDefaultInstance(), connection, registry);
    final SimpleRootExec exec = new SimpleRootExec(ImplCreator.getExec(context, (FragmentRoot) plan.getSortedOperators(false).iterator().next()));
    boolean oneIsOne = false;
    int size = 0;
    final int[] sizes = { 1, 2, 0, 6 };
    while (exec.next()) {
        final IntVector c1 = exec.getValueVectorById(new SchemaPath("cnt", ExpressionPosition.UNKNOWN), IntVector.class);
        final BitVector c2 = exec.getValueVectorById(new SchemaPath("has_min", ExpressionPosition.UNKNOWN), BitVector.class);
        for (int i = 0; i < exec.getRecordCount(); i++) {
            final int curSize = sizes[size % sizes.length];
            assertEquals(curSize, c1.getAccessor().get(i));
            switch(curSize) {
                case 1:
                    assertEquals(oneIsOne, 1 == c2.getAccessor().get(i));
                    oneIsOne = !oneIsOne;
                    break;
                case 2:
                    assertEquals(1, c2.getAccessor().get(i));
                    break;
                case 0:
                    assertEquals(0, c2.getAccessor().get(i));
                    break;
                case 6:
                    assertEquals(1, c2.getAccessor().get(i));
                    break;
            }
            size++;
        }
    }
    if (context.getExecutorState().getFailureCause() != null) {
        throw context.getExecutorState().getFailureCause();
    }
    assertTrue(!context.getExecutorState().isFailed());
}
Also used : DrillbitContext(org.apache.drill.exec.server.DrillbitContext) BitVector(org.apache.drill.exec.vector.BitVector) PhysicalPlan(org.apache.drill.exec.physical.PhysicalPlan) IntVector(org.apache.drill.exec.vector.IntVector) PhysicalPlanReader(org.apache.drill.exec.planner.PhysicalPlanReader) FragmentContextImpl(org.apache.drill.exec.ops.FragmentContextImpl) FragmentRoot(org.apache.drill.exec.physical.base.FragmentRoot) SimpleRootExec(org.apache.drill.exec.physical.impl.SimpleRootExec) SchemaPath(org.apache.drill.common.expression.SchemaPath) UserClientConnection(org.apache.drill.exec.rpc.UserClientConnection) FunctionImplementationRegistry(org.apache.drill.exec.expr.fn.FunctionImplementationRegistry) ExecTest(org.apache.drill.exec.ExecTest) OperatorTest(org.apache.drill.categories.OperatorTest) Test(org.junit.Test)

Aggregations

UserClientConnection (org.apache.drill.exec.rpc.UserClientConnection)78 DrillbitContext (org.apache.drill.exec.server.DrillbitContext)78 FunctionImplementationRegistry (org.apache.drill.exec.expr.fn.FunctionImplementationRegistry)72 FragmentContextImpl (org.apache.drill.exec.ops.FragmentContextImpl)72 PhysicalPlan (org.apache.drill.exec.physical.PhysicalPlan)70 FragmentRoot (org.apache.drill.exec.physical.base.FragmentRoot)70 Test (org.junit.Test)66 PhysicalPlanReader (org.apache.drill.exec.planner.PhysicalPlanReader)62 SimpleRootExec (org.apache.drill.exec.physical.impl.SimpleRootExec)42 OperatorTest (org.apache.drill.categories.OperatorTest)36 ExecTest (org.apache.drill.exec.ExecTest)34 SchemaPath (org.apache.drill.common.expression.SchemaPath)32 ValueVector (org.apache.drill.exec.vector.ValueVector)14 Ignore (org.junit.Ignore)14 IntVector (org.apache.drill.exec.vector.IntVector)12 SlowTest (org.apache.drill.categories.SlowTest)10 BigIntVector (org.apache.drill.exec.vector.BigIntVector)10 BigIntHolder (org.apache.drill.exec.expr.holders.BigIntHolder)6 StoragePluginRegistryImpl (org.apache.drill.exec.store.StoragePluginRegistryImpl)6 Configuration (org.apache.hadoop.conf.Configuration)6