Examples with ColumnStatistics org.apache.hadoop.hive.metastore.api.ColumnStatistics used on opensource projects

Example 86 with ColumnStatistics

use of org.apache.hadoop.hive.metastore.api.ColumnStatistics in project hive by apache.

the class SharedCache method populateTableInCache.

public void populateTableInCache(Table table, ColumnStatistics tableColStats, List<Partition> partitions, List<ColumnStatistics> partitionColStats, AggrStats aggrStatsAllPartitions, AggrStats aggrStatsAllButDefaultPartition) {
    String dbName = StringUtils.normalizeIdentifier(table.getDbName());
    String tableName = StringUtils.normalizeIdentifier(table.getTableName());
    // 1. Don't add tables that were deleted while we were preparing list for prewarm
    if (tablesDeletedDuringPrewarm.contains(CacheUtils.buildTableCacheKey(dbName, tableName))) {
        return;
    }
    TableWrapper tblWrapper = createTableWrapper(dbName, tableName, table);
    if (!table.isSetPartitionKeys() && (tableColStats != null)) {
        tblWrapper.updateTableColStats(tableColStats.getStatsObj());
    } else {
        if (partitions != null) {
            tblWrapper.cachePartitions(partitions, this);
        }
        if (partitionColStats != null) {
            for (ColumnStatistics cs : partitionColStats) {
                List<String> partVal;
                try {
                    partVal = Warehouse.makeValsFromName(cs.getStatsDesc().getPartName(), null);
                    List<ColumnStatisticsObj> colStats = cs.getStatsObj();
                    tblWrapper.updatePartitionColStats(partVal, colStats);
                } catch (MetaException e) {
                    LOG.debug("Unable to cache partition column stats for table: " + tableName, e);
                }
            }
        }
        tblWrapper.cacheAggrPartitionColStats(aggrStatsAllPartitions, aggrStatsAllButDefaultPartition);
    }
    try {
        cacheLock.writeLock().lock();
        // 2. Skip overwriting exisiting table object
        // (which is present because it was added after prewarm started)
        tableCache.putIfAbsent(CacheUtils.buildTableCacheKey(dbName, tableName), tblWrapper);
    } finally {
        cacheLock.writeLock().unlock();
    }
}

Example 87 with ColumnStatistics

use of org.apache.hadoop.hive.metastore.api.ColumnStatistics in project hive by apache.

the class VerifyingObjectStore method getTableColumnStatistics.

@Override
public ColumnStatistics getTableColumnStatistics(String dbName, String tableName, List<String> colNames) throws MetaException, NoSuchObjectException {
    ColumnStatistics sqlResult = getTableColumnStatisticsInternal(dbName, tableName, colNames, true, false);
    ColumnStatistics jdoResult = getTableColumnStatisticsInternal(dbName, tableName, colNames, false, true);
    verifyObjects(sqlResult, jdoResult, ColumnStatistics.class);
    return sqlResult;
}

Example 88 with ColumnStatistics

use of org.apache.hadoop.hive.metastore.api.ColumnStatistics in project hive by apache.

the class TestCachedStore method testPartitionAggrStatsBitVector.

@Test
public void testPartitionAggrStatsBitVector() throws Exception {
    String dbName = "testTableColStatsOps2";
    String tblName = "tbl2";
    String colName = "f1";
    Database db = new Database(dbName, null, "some_location", null);
    cachedStore.createDatabase(db);
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema(colName, "int", null));
    List<FieldSchema> partCols = new ArrayList<>();
    partCols.add(new FieldSchema("col", "int", null));
    StorageDescriptor sd = new StorageDescriptor(cols, null, "input", "output", false, 0, new SerDeInfo("serde", "seriallib", new HashMap<>()), null, null, null);
    Table tbl = new Table(tblName, dbName, null, 0, 0, 0, sd, partCols, new HashMap<>(), null, null, TableType.MANAGED_TABLE.toString());
    cachedStore.createTable(tbl);
    List<String> partVals1 = new ArrayList<>();
    partVals1.add("1");
    List<String> partVals2 = new ArrayList<>();
    partVals2.add("2");
    Partition ptn1 = new Partition(partVals1, dbName, tblName, 0, 0, sd, new HashMap<>());
    cachedStore.addPartition(ptn1);
    Partition ptn2 = new Partition(partVals2, dbName, tblName, 0, 0, sd, new HashMap<>());
    cachedStore.addPartition(ptn2);
    ColumnStatistics stats = new ColumnStatistics();
    ColumnStatisticsDesc statsDesc = new ColumnStatisticsDesc(true, dbName, tblName);
    statsDesc.setPartName("col");
    List<ColumnStatisticsObj> colStatObjs = new ArrayList<>();
    ColumnStatisticsData data = new ColumnStatisticsData();
    ColumnStatisticsObj colStats = new ColumnStatisticsObj(colName, "int", data);
    LongColumnStatsDataInspector longStats = new LongColumnStatsDataInspector();
    longStats.setLowValue(0);
    longStats.setHighValue(100);
    longStats.setNumNulls(50);
    longStats.setNumDVs(30);
    HyperLogLog hll = HyperLogLog.builder().build();
    hll.addLong(1);
    hll.addLong(2);
    hll.addLong(3);
    longStats.setBitVectors(hll.serialize());
    data.setLongStats(longStats);
    colStatObjs.add(colStats);
    stats.setStatsDesc(statsDesc);
    stats.setStatsObj(colStatObjs);
    cachedStore.updatePartitionColumnStatistics(stats.deepCopy(), partVals1);
    longStats.setNumDVs(40);
    hll = HyperLogLog.builder().build();
    hll.addLong(2);
    hll.addLong(3);
    hll.addLong(4);
    hll.addLong(5);
    longStats.setBitVectors(hll.serialize());
    cachedStore.updatePartitionColumnStatistics(stats.deepCopy(), partVals2);
    List<String> colNames = new ArrayList<>();
    colNames.add(colName);
    List<String> aggrPartVals = new ArrayList<>();
    aggrPartVals.add("1");
    aggrPartVals.add("2");
    AggrStats aggrStats = cachedStore.get_aggr_stats_for(dbName, tblName, aggrPartVals, colNames);
    Assert.assertEquals(aggrStats.getColStats().get(0).getStatsData().getLongStats().getNumNulls(), 100);
    Assert.assertEquals(aggrStats.getColStats().get(0).getStatsData().getLongStats().getNumDVs(), 5);
    aggrStats = cachedStore.get_aggr_stats_for(dbName, tblName, aggrPartVals, colNames);
    Assert.assertEquals(aggrStats.getColStats().get(0).getStatsData().getLongStats().getNumNulls(), 100);
    Assert.assertEquals(aggrStats.getColStats().get(0).getStatsData().getLongStats().getNumDVs(), 5);
}

Example 89 with ColumnStatistics

use of org.apache.hadoop.hive.metastore.api.ColumnStatistics in project hive by apache.

the class TestCachedStore method testTableColStatsOps.

// @Test
public void testTableColStatsOps() throws Exception {
    // Add a db via ObjectStore
    String dbName = "testTableColStatsOps";
    String dbOwner = "user1";
    Database db = createTestDb(dbName, dbOwner);
    objectStore.createDatabase(db);
    db = objectStore.getDatabase(dbName);
    // Add a table via ObjectStore
    final String tblName = "tbl";
    final String tblOwner = "user1";
    final FieldSchema col1 = new FieldSchema("col1", "int", "integer column");
    // Stats values for col1
    long col1LowVal = 5;
    long col1HighVal = 500;
    long col1Nulls = 10;
    long col1DV = 20;
    final FieldSchema col2 = new FieldSchema("col2", "string", "string column");
    // Stats values for col2
    long col2MaxColLen = 100;
    double col2AvgColLen = 45.5;
    long col2Nulls = 5;
    long col2DV = 40;
    final FieldSchema col3 = new FieldSchema("col3", "boolean", "boolean column");
    // Stats values for col3
    long col3NumTrues = 100;
    long col3NumFalses = 30;
    long col3Nulls = 10;
    final List<FieldSchema> cols = new ArrayList<>();
    cols.add(col1);
    cols.add(col2);
    cols.add(col3);
    FieldSchema ptnCol1 = new FieldSchema("part1", "string", "string partition column");
    List<FieldSchema> ptnCols = new ArrayList<FieldSchema>();
    ptnCols.add(ptnCol1);
    Table tbl = createTestTbl(dbName, tblName, tblOwner, cols, ptnCols);
    objectStore.createTable(tbl);
    tbl = objectStore.getTable(dbName, tblName);
    // Add ColumnStatistics for tbl to metastore DB via ObjectStore
    ColumnStatistics stats = new ColumnStatistics();
    ColumnStatisticsDesc statsDesc = new ColumnStatisticsDesc(true, dbName, tblName);
    List<ColumnStatisticsObj> colStatObjs = new ArrayList<>();
    // Col1
    ColumnStatisticsData data1 = new ColumnStatisticsData();
    ColumnStatisticsObj col1Stats = new ColumnStatisticsObj(col1.getName(), col1.getType(), data1);
    LongColumnStatsDataInspector longStats = new LongColumnStatsDataInspector();
    longStats.setLowValue(col1LowVal);
    longStats.setHighValue(col1HighVal);
    longStats.setNumNulls(col1Nulls);
    longStats.setNumDVs(col1DV);
    data1.setLongStats(longStats);
    colStatObjs.add(col1Stats);
    // Col2
    ColumnStatisticsData data2 = new ColumnStatisticsData();
    ColumnStatisticsObj col2Stats = new ColumnStatisticsObj(col2.getName(), col2.getType(), data2);
    StringColumnStatsDataInspector stringStats = new StringColumnStatsDataInspector();
    stringStats.setMaxColLen(col2MaxColLen);
    stringStats.setAvgColLen(col2AvgColLen);
    stringStats.setNumNulls(col2Nulls);
    stringStats.setNumDVs(col2DV);
    data2.setStringStats(stringStats);
    colStatObjs.add(col2Stats);
    // Col3
    ColumnStatisticsData data3 = new ColumnStatisticsData();
    ColumnStatisticsObj col3Stats = new ColumnStatisticsObj(col3.getName(), col3.getType(), data3);
    BooleanColumnStatsData boolStats = new BooleanColumnStatsData();
    boolStats.setNumTrues(col3NumTrues);
    boolStats.setNumFalses(col3NumFalses);
    boolStats.setNumNulls(col3Nulls);
    data3.setBooleanStats(boolStats);
    colStatObjs.add(col3Stats);
    stats.setStatsDesc(statsDesc);
    stats.setStatsObj(colStatObjs);
    // Save to DB
    objectStore.updateTableColumnStatistics(stats);
    // Prewarm CachedStore
    CachedStore.setCachePrewarmedState(false);
    CachedStore.prewarm(objectStore);
    // Read table stats via CachedStore
    ColumnStatistics newStats = cachedStore.getTableColumnStatistics(dbName, tblName, Arrays.asList(col1.getName(), col2.getName(), col3.getName()));
    Assert.assertEquals(stats, newStats);
    // Clean up
    objectStore.dropTable(dbName, tblName);
    objectStore.dropDatabase(dbName);
    sharedCache.getDatabaseCache().clear();
    sharedCache.getTableCache().clear();
    sharedCache.getSdCache().clear();
}

Example 90 with ColumnStatistics

use of org.apache.hadoop.hive.metastore.api.ColumnStatistics in project hive by apache.

the class TestCachedStore method testPartitionAggrStats.

@Test
public void testPartitionAggrStats() throws Exception {
    String dbName = "testTableColStatsOps1";
    String tblName = "tbl1";
    String colName = "f1";
    Database db = new Database(dbName, null, "some_location", null);
    cachedStore.createDatabase(db);
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema(colName, "int", null));
    List<FieldSchema> partCols = new ArrayList<>();
    partCols.add(new FieldSchema("col", "int", null));
    StorageDescriptor sd = new StorageDescriptor(cols, null, "input", "output", false, 0, new SerDeInfo("serde", "seriallib", new HashMap<>()), null, null, null);
    Table tbl = new Table(tblName, dbName, null, 0, 0, 0, sd, partCols, new HashMap<>(), null, null, TableType.MANAGED_TABLE.toString());
    cachedStore.createTable(tbl);
    List<String> partVals1 = new ArrayList<>();
    partVals1.add("1");
    List<String> partVals2 = new ArrayList<>();
    partVals2.add("2");
    Partition ptn1 = new Partition(partVals1, dbName, tblName, 0, 0, sd, new HashMap<>());
    cachedStore.addPartition(ptn1);
    Partition ptn2 = new Partition(partVals2, dbName, tblName, 0, 0, sd, new HashMap<>());
    cachedStore.addPartition(ptn2);
    ColumnStatistics stats = new ColumnStatistics();
    ColumnStatisticsDesc statsDesc = new ColumnStatisticsDesc(true, dbName, tblName);
    statsDesc.setPartName("col");
    List<ColumnStatisticsObj> colStatObjs = new ArrayList<>();
    ColumnStatisticsData data = new ColumnStatisticsData();
    ColumnStatisticsObj colStats = new ColumnStatisticsObj(colName, "int", data);
    LongColumnStatsDataInspector longStats = new LongColumnStatsDataInspector();
    longStats.setLowValue(0);
    longStats.setHighValue(100);
    longStats.setNumNulls(50);
    longStats.setNumDVs(30);
    data.setLongStats(longStats);
    colStatObjs.add(colStats);
    stats.setStatsDesc(statsDesc);
    stats.setStatsObj(colStatObjs);
    cachedStore.updatePartitionColumnStatistics(stats.deepCopy(), partVals1);
    longStats.setNumDVs(40);
    cachedStore.updatePartitionColumnStatistics(stats.deepCopy(), partVals2);
    List<String> colNames = new ArrayList<>();
    colNames.add(colName);
    List<String> aggrPartVals = new ArrayList<>();
    aggrPartVals.add("1");
    aggrPartVals.add("2");
    AggrStats aggrStats = cachedStore.get_aggr_stats_for(dbName, tblName, aggrPartVals, colNames);
    Assert.assertEquals(aggrStats.getColStats().get(0).getStatsData().getLongStats().getNumNulls(), 100);
    Assert.assertEquals(aggrStats.getColStats().get(0).getStatsData().getLongStats().getNumDVs(), 40);
    aggrStats = cachedStore.get_aggr_stats_for(dbName, tblName, aggrPartVals, colNames);
    Assert.assertEquals(aggrStats.getColStats().get(0).getStatsData().getLongStats().getNumNulls(), 100);
    Assert.assertEquals(aggrStats.getColStats().get(0).getStatsData().getLongStats().getNumDVs(), 40);
}