Examples with SegmentMetadata org.apache.drill.metastore.metadata.SegmentMetadata used on opensource projects

Example 16 with SegmentMetadata

use of org.apache.drill.metastore.metadata.SegmentMetadata in project drill by apache.

the class TestMetastoreCommands method testIncrementalAnalyzeRemovedParentSegment.

@Test
public void testIncrementalAnalyzeRemovedParentSegment() throws Exception {
    String tableName = "multilevel/parquetRemovedParent";
    File table = dirTestWatcher.copyResourceToTestTmp(Paths.get("multilevel/parquet"), Paths.get(tableName));
    TableInfo tableInfo = getTableInfo(tableName, "tmp");
    BaseTableMetadata expectedTableMetadata = getBaseTableMetadata(tableInfo, table);
    try {
        dirTestWatcher.copyResourceToTestTmp(Paths.get("multilevel/parquet", "1994"), Paths.get(tableName, "1993"));
        testBuilder().sqlQuery("ANALYZE TABLE dfs.tmp.`%s` REFRESH METADATA", tableName).unOrdered().baselineColumns("ok", "summary").baselineValues(true, String.format("Collected / refreshed metadata for table [dfs.tmp.%s]", tableName)).go();
        List<SegmentMetadata> segmentMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().segmentsMetadataByMetadataKey(tableInfo, null, null);
        assertEquals(20, segmentMetadata.size());
        List<FileMetadata> filesMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().filesMetadata(tableInfo, null, null);
        assertEquals(16, filesMetadata.size());
        List<RowGroupMetadata> rowGroupsMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().rowGroupsMetadata(tableInfo, null, (String) null);
        assertEquals(16, rowGroupsMetadata.size());
        FileUtils.deleteQuietly(new File(table, "1993"));
        testBuilder().sqlQuery("ANALYZE TABLE dfs.tmp.`%s` REFRESH METADATA", tableName).unOrdered().baselineColumns("ok", "summary").baselineValues(true, String.format("Collected / refreshed metadata for table [dfs.tmp.%s]", tableName)).go();
        BaseTableMetadata actualTableMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().tableMetadata(tableInfo);
        assertEquals(expectedTableMetadata, actualTableMetadata);
        segmentMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().segmentsMetadataByMetadataKey(tableInfo, null, null);
        assertEquals(15, segmentMetadata.size());
        filesMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().filesMetadata(tableInfo, null, null);
        assertEquals(12, filesMetadata.size());
        rowGroupsMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().rowGroupsMetadata(tableInfo, null, (String) null);
        assertEquals(12, rowGroupsMetadata.size());
    } finally {
        run("analyze table dfs.tmp.`%s` drop metadata if exists", tableName);
        FileUtils.deleteQuietly(table);
    }
}

Example 17 with SegmentMetadata

use of org.apache.drill.metastore.metadata.SegmentMetadata in project drill by apache.

the class TestMetastoreCommands method testIncrementalAnalyzeUnchangedTable.

@Test
public void testIncrementalAnalyzeUnchangedTable() throws Exception {
    String tableName = "multilevel/parquetUnchanged";
    File table = dirTestWatcher.copyResourceToTestTmp(Paths.get("multilevel/parquet"), Paths.get(tableName));
    TableInfo tableInfo = getTableInfo(tableName, "tmp");
    long lastModifiedTime = getMaxLastModified(table);
    try {
        testBuilder().sqlQuery("ANALYZE TABLE dfs.tmp.`%s` REFRESH METADATA", tableName).unOrdered().baselineColumns("ok", "summary").baselineValues(true, String.format("Collected / refreshed metadata for table [dfs.tmp.%s]", tableName)).go();
        List<SegmentMetadata> segmentMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().segmentsMetadataByMetadataKey(tableInfo, null, null);
        assertEquals(15, segmentMetadata.size());
        testBuilder().sqlQuery("ANALYZE TABLE dfs.tmp.`%s` REFRESH METADATA", tableName).unOrdered().baselineColumns("ok", "summary").baselineValues(false, "Table metadata is up to date, analyze wasn't performed.").go();
        segmentMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().segmentsMetadataByMetadataKey(tableInfo, null, null);
        assertEquals(15, segmentMetadata.size());
        long postAnalyzeLastModifiedTime = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().metastoreTableInfo(tableInfo).lastModifiedTime();
        assertEquals(lastModifiedTime, postAnalyzeLastModifiedTime);
    } finally {
        run("analyze table dfs.tmp.`%s` drop metadata if exists", tableName);
        FileUtils.deleteQuietly(table);
    }
}

Example 18 with SegmentMetadata

use of org.apache.drill.metastore.metadata.SegmentMetadata in project drill by apache.

the class TestMetastoreCommands method testIncrementalAnalyzeNewChildSegment.

@Test
public void testIncrementalAnalyzeNewChildSegment() throws Exception {
    String tableName = "multilevel/parquetNewChildSegment";
    File table = dirTestWatcher.copyResourceToTestTmp(Paths.get("multilevel/parquet"), Paths.get(tableName));
    Path tablePath = new Path(table.toURI().getPath());
    TableInfo tableInfo = getTableInfo(tableName, "tmp");
    // updates statistics values due to new segment
    Map<SchemaPath, ColumnStatistics<?>> updatedStatistics = new HashMap<>(TABLE_COLUMN_STATISTICS);
    updatedStatistics.replaceAll((logicalExpressions, columnStatistics) -> columnStatistics.cloneWith(new ColumnStatistics<>(Arrays.asList(new StatisticsHolder<>(130L, TableStatisticsKind.ROW_COUNT), new StatisticsHolder<>(130L, ColumnStatisticsKind.NON_NULL_VALUES_COUNT)))));
    updatedStatistics.computeIfPresent(SchemaPath.getSimplePath("dir1"), (logicalExpressions, columnStatistics) -> columnStatistics.cloneWith(new ColumnStatistics<>(Collections.singletonList(new StatisticsHolder<>("Q5", ColumnStatisticsKind.MAX_VALUE)))));
    BaseTableMetadata expectedTableMetadata = BaseTableMetadata.builder().tableInfo(tableInfo).metadataInfo(TABLE_META_INFO).schema(SCHEMA).location(tablePath).columnsStatistics(updatedStatistics).metadataStatistics(Arrays.asList(new StatisticsHolder<>(130L, TableStatisticsKind.ROW_COUNT), new StatisticsHolder<>(MetadataType.ALL, TableStatisticsKind.ANALYZE_METADATA_LEVEL))).partitionKeys(Collections.emptyMap()).lastModifiedTime(getMaxLastModified(table)).build();
    try {
        testBuilder().sqlQuery("ANALYZE TABLE dfs.tmp.`%s` REFRESH METADATA", tableName).unOrdered().baselineColumns("ok", "summary").baselineValues(true, String.format("Collected / refreshed metadata for table [dfs.tmp.%s]", tableName)).go();
        List<SegmentMetadata> segmentMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().segmentsMetadataByMetadataKey(tableInfo, null, null);
        assertEquals(15, segmentMetadata.size());
        dirTestWatcher.copyResourceToTestTmp(Paths.get("multilevel", "parquet", "1994", "Q4"), Paths.get(tableName, "1994", "Q5"));
        testBuilder().sqlQuery("ANALYZE TABLE dfs.tmp.`%s` REFRESH METADATA", tableName).unOrdered().baselineColumns("ok", "summary").baselineValues(true, String.format("Collected / refreshed metadata for table [dfs.tmp.%s]", tableName)).go();
        BaseTableMetadata actualTableMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().tableMetadata(tableInfo);
        assertEquals(expectedTableMetadata, actualTableMetadata);
        segmentMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().segmentsMetadataByMetadataKey(tableInfo, null, null);
        assertEquals(16, segmentMetadata.size());
    } finally {
        run("analyze table dfs.tmp.`%s` drop metadata if exists", tableName);
        FileUtils.deleteQuietly(table);
    }
}

Example 19 with SegmentMetadata

use of org.apache.drill.metastore.metadata.SegmentMetadata in project drill by apache.

the class TestMetastoreCommands method testIncrementalAnalyzeNewParentSegment.

@Test
public void testIncrementalAnalyzeNewParentSegment() throws Exception {
    String tableName = "multilevel/parquetNewParentSegment";
    File table = dirTestWatcher.copyResourceToTestTmp(Paths.get("multilevel/parquet"), Paths.get(tableName));
    Path tablePath = new Path(table.toURI().getPath());
    TableInfo tableInfo = getTableInfo(tableName, "tmp");
    // updates statistics values due to new segment
    Map<SchemaPath, ColumnStatistics<?>> updatedStatistics = new HashMap<>(TABLE_COLUMN_STATISTICS);
    updatedStatistics.replaceAll((logicalExpressions, columnStatistics) -> columnStatistics.cloneWith(new ColumnStatistics<>(Arrays.asList(new StatisticsHolder<>(160L, TableStatisticsKind.ROW_COUNT), new StatisticsHolder<>(160L, ColumnStatisticsKind.NON_NULL_VALUES_COUNT)))));
    updatedStatistics.computeIfPresent(SchemaPath.getSimplePath("dir0"), (logicalExpressions, columnStatistics) -> columnStatistics.cloneWith(new ColumnStatistics<>(Collections.singletonList(new StatisticsHolder<>("1993", ColumnStatisticsKind.MIN_VALUE)))));
    BaseTableMetadata expectedTableMetadata = BaseTableMetadata.builder().tableInfo(tableInfo).metadataInfo(TABLE_META_INFO).schema(SCHEMA).location(tablePath).columnsStatistics(updatedStatistics).metadataStatistics(Arrays.asList(new StatisticsHolder<>(160L, TableStatisticsKind.ROW_COUNT), new StatisticsHolder<>(MetadataType.ALL, TableStatisticsKind.ANALYZE_METADATA_LEVEL))).partitionKeys(Collections.emptyMap()).lastModifiedTime(getMaxLastModified(table)).build();
    try {
        assertEquals(0, cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().segmentsMetadataByMetadataKey(tableInfo, null, null).size());
        testBuilder().sqlQuery("ANALYZE TABLE dfs.tmp.`%s` REFRESH METADATA", tableName).unOrdered().baselineColumns("ok", "summary").baselineValues(true, String.format("Collected / refreshed metadata for table [dfs.tmp.%s]", tableName)).go();
        List<SegmentMetadata> segmentMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().segmentsMetadataByMetadataKey(tableInfo, null, null);
        assertEquals(15, segmentMetadata.size());
        dirTestWatcher.copyResourceToTestTmp(Paths.get("multilevel/parquet", "1994"), Paths.get(tableName, "1993"));
        testBuilder().sqlQuery("ANALYZE TABLE dfs.tmp.`%s` REFRESH METADATA", tableName).unOrdered().baselineColumns("ok", "summary").baselineValues(true, String.format("Collected / refreshed metadata for table [dfs.tmp.%s]", tableName)).go();
        BaseTableMetadata actualTableMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().tableMetadata(tableInfo);
        assertEquals(expectedTableMetadata, actualTableMetadata);
        segmentMetadata = cluster.drillbit().getContext().getMetastoreRegistry().get().tables().basicRequests().segmentsMetadataByMetadataKey(tableInfo, null, null);
        assertEquals(20, segmentMetadata.size());
    } finally {
        run("analyze table dfs.tmp.`%s` drop metadata if exists", tableName);
        FileUtils.deleteQuietly(table);
    }
}

Example 20 with SegmentMetadata

use of org.apache.drill.metastore.metadata.SegmentMetadata in project drill by apache.

the class MetadataHandlerBatch method writeMetadata.

private <T extends BaseMetadata & LocationProvider> VectorContainer writeMetadata(List<T> metadataList) {
    BaseMetadata firstElement = metadataList.iterator().next();
    ResultSetLoader resultSetLoader = getResultSetLoaderForMetadata(firstElement);
    resultSetLoader.startBatch();
    RowSetLoader rowWriter = resultSetLoader.writer();
    Iterator<T> segmentsIterator = metadataList.iterator();
    while (!rowWriter.isFull() && segmentsIterator.hasNext()) {
        T metadata = segmentsIterator.next();
        metadataToHandle.remove(metadata.getMetadataInfo().identifier());
        List<Object> arguments = new ArrayList<>();
        // adds required segment names to the arguments
        arguments.add(metadata.getPath().toUri().getPath());
        Collections.addAll(arguments, Arrays.copyOf(MetadataIdentifierUtils.getValuesFromMetadataIdentifier(metadata.getMetadataInfo().identifier()), popConfig.getContext().segmentColumns().size()));
        // adds column statistics values assuming that they are sorted in alphabetic order
        // (see getResultSetLoaderForMetadata() method)
        metadata.getColumnsStatistics().entrySet().stream().sorted(Comparator.comparing(e -> e.getKey().toExpr())).map(Map.Entry::getValue).flatMap(columnStatistics -> AnalyzeColumnUtils.COLUMN_STATISTICS_FUNCTIONS.keySet().stream().map(columnStatistics::get)).forEach(arguments::add);
        AnalyzeColumnUtils.META_STATISTICS_FUNCTIONS.keySet().stream().map(metadata::getStatistic).forEach(arguments::add);
        // collectedMap field value
        arguments.add(new Object[] {});
        if (metadataType == MetadataType.SEGMENT) {
            arguments.add(((SegmentMetadata) metadata).getLocations().stream().map(path -> path.toUri().getPath()).toArray(String[]::new));
        }
        if (metadataType == MetadataType.ROW_GROUP) {
            arguments.add(String.valueOf(((RowGroupMetadata) metadata).getRowGroupIndex()));
            arguments.add(Long.toString(metadata.getStatistic(() -> ExactStatisticsConstants.START)));
            arguments.add(Long.toString(metadata.getStatistic(() -> ExactStatisticsConstants.LENGTH)));
        }
        arguments.add(metadata.getSchema().jsonString());
        arguments.add(String.valueOf(metadata.getLastModifiedTime()));
        arguments.add(metadataType.name());
        rowWriter.addRow(arguments.toArray());
    }
    return resultSetLoader.harvest();
}