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

Example 6 with ColumnStatistics

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

the class TestHBaseAggregateStatsCache method allWithStats.

// Do to limitations in the Mock infrastructure we use for HBase testing we can only test
// this for a single column table and we can't really test hits in hbase, only in memory or
// build from scratch.  But it's still useful to cover many bugs.  More in depth testing with
// multiple columns and with HBase hits is done in TestHBaseAggrStatsCacheIntegration.
@Test
public void allWithStats() throws Exception {
    String dbName = "default";
    String tableName = "hit";
    List<String> partVals1 = Arrays.asList("today");
    List<String> partVals2 = Arrays.asList("yesterday");
    long now = System.currentTimeMillis();
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema("col1", "boolean", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, Collections.<String, String>emptyMap());
    List<FieldSchema> partCols = new ArrayList<>();
    partCols.add(new FieldSchema("ds", "string", ""));
    Table table = new Table(tableName, dbName, "me", (int) now, (int) now, 0, sd, partCols, Collections.<String, String>emptyMap(), null, null, null);
    store.createTable(table);
    for (List<String> partVals : Arrays.asList(partVals1, partVals2)) {
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/default/hit/ds=" + partVals.get(0));
        Partition part = new Partition(partVals, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
        store.addPartition(part);
        ColumnStatistics cs = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
        desc.setLastAnalyzed(now);
        desc.setPartName("ds=" + partVals.get(0));
        cs.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName("col1");
        obj.setColType("boolean");
        ColumnStatisticsData data = new ColumnStatisticsData();
        BooleanColumnStatsData bcsd = new BooleanColumnStatsData();
        bcsd.setNumFalses(10);
        bcsd.setNumTrues(20);
        bcsd.setNumNulls(30);
        data.setBooleanStats(bcsd);
        obj.setStatsData(data);
        cs.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(cs, partVals);
    }
    Checker statChecker = new Checker() {

        @Override
        public void checkStats(AggrStats aggrStats) throws Exception {
            Assert.assertEquals(2, aggrStats.getPartsFound());
            Assert.assertEquals(1, aggrStats.getColStatsSize());
            ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
            Assert.assertEquals("col1", cso.getColName());
            Assert.assertEquals("boolean", cso.getColType());
            BooleanColumnStatsData bcsd = cso.getStatsData().getBooleanStats();
            Assert.assertEquals(20, bcsd.getNumFalses());
            Assert.assertEquals(40, bcsd.getNumTrues());
            Assert.assertEquals(60, bcsd.getNumNulls());
        }
    };
    AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=yesterday"), Arrays.asList("col1"));
    statChecker.checkStats(aggrStats);
    // Check that we had to build it from the stats
    Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
    Assert.assertEquals(1, store.backdoor().getStatsCache().totalGets.getCnt());
    Assert.assertEquals(1, store.backdoor().getStatsCache().misses.getCnt());
    // Call again, this time it should come from memory.  Also, reverse the name order this time
    // to assure that we still hit.
    aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=yesterday", "ds=today"), Arrays.asList("col1"));
    statChecker.checkStats(aggrStats);
    Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
    Assert.assertEquals(2, store.backdoor().getStatsCache().totalGets.getCnt());
    Assert.assertEquals(1, store.backdoor().getStatsCache().misses.getCnt());
}

Example 7 with ColumnStatistics

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

the class TestHBaseAggregateStatsCache method someNonexistentPartitions.

@Test
public void someNonexistentPartitions() throws Exception {
    String dbName = "default";
    String tableName = "snp";
    List<String> partVals1 = Arrays.asList("today");
    List<String> partVals2 = Arrays.asList("yesterday");
    long now = System.currentTimeMillis();
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema("col1", "boolean", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, Collections.<String, String>emptyMap());
    List<FieldSchema> partCols = new ArrayList<>();
    partCols.add(new FieldSchema("ds", "string", ""));
    Table table = new Table(tableName, dbName, "me", (int) now, (int) now, 0, sd, partCols, Collections.<String, String>emptyMap(), null, null, null);
    store.createTable(table);
    StorageDescriptor psd = new StorageDescriptor(sd);
    psd.setLocation("file:/tmp/default/hit/ds=" + partVals1.get(0));
    Partition part = new Partition(partVals1, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
    store.addPartition(part);
    ColumnStatistics cs = new ColumnStatistics();
    ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
    desc.setLastAnalyzed(now);
    desc.setPartName("ds=" + partVals1.get(0));
    cs.setStatsDesc(desc);
    ColumnStatisticsObj obj = new ColumnStatisticsObj();
    obj.setColName("col1");
    obj.setColType("double");
    ColumnStatisticsData data = new ColumnStatisticsData();
    DoubleColumnStatsData dcsd = new DoubleColumnStatsData();
    dcsd.setHighValue(1000.2342343);
    dcsd.setLowValue(-20.1234213423);
    dcsd.setNumNulls(30);
    dcsd.setNumDVs(12342);
    data.setDoubleStats(dcsd);
    obj.setStatsData(data);
    cs.addToStatsObj(obj);
    store.updatePartitionColumnStatistics(cs, partVals1);
    Checker statChecker = new Checker() {

        @Override
        public void checkStats(AggrStats aggrStats) throws Exception {
            Assert.assertEquals(1, aggrStats.getPartsFound());
            Assert.assertEquals(1, aggrStats.getColStatsSize());
            ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
            Assert.assertEquals("col1", cso.getColName());
            Assert.assertEquals("double", cso.getColType());
            DoubleColumnStatsData dcsd = cso.getStatsData().getDoubleStats();
            Assert.assertEquals(1000.23, dcsd.getHighValue(), 0.01);
            Assert.assertEquals(-20.12, dcsd.getLowValue(), 0.01);
            Assert.assertEquals(30, dcsd.getNumNulls());
            Assert.assertEquals(12342, dcsd.getNumDVs());
        }
    };
    AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=yesterday"), Arrays.asList("col1"));
    statChecker.checkStats(aggrStats);
    // Check that we had to build it from the stats
    Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
    Assert.assertEquals(1, store.backdoor().getStatsCache().totalGets.getCnt());
    Assert.assertEquals(1, store.backdoor().getStatsCache().misses.getCnt());
    // Call again, this time it should come from memory.  Also, reverse the name order this time
    // to assure that we still hit.
    aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=yesterday", "ds=today"), Arrays.asList("col1"));
    statChecker.checkStats(aggrStats);
    Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
    Assert.assertEquals(2, store.backdoor().getStatsCache().totalGets.getCnt());
    Assert.assertEquals(1, store.backdoor().getStatsCache().misses.getCnt());
}

Example 8 with ColumnStatistics

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

the class TestHBaseAggregateStatsExtrapolation method allPartitionsHaveBitVectorStatusDecimal.

@Test
public void allPartitionsHaveBitVectorStatusDecimal() throws Exception {
    String dbName = "default";
    String tableName = "snp";
    long now = System.currentTimeMillis();
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema("col1_decimal", "decimal", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, Collections.<String, String>emptyMap());
    List<FieldSchema> partCols = new ArrayList<>();
    partCols.add(new FieldSchema("ds", "string", ""));
    Table table = new Table(tableName, dbName, "me", (int) now, (int) now, 0, sd, partCols, Collections.<String, String>emptyMap(), null, null, null);
    store.createTable(table);
    List<List<String>> partVals = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        List<String> partVal = Arrays.asList("" + i);
        partVals.add(partVal);
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/default/hit/ds=" + partVal);
        Partition part = new Partition(partVal, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
        store.addPartition(part);
        ColumnStatistics cs = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
        desc.setLastAnalyzed(now);
        desc.setPartName("ds=" + partVal);
        cs.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName("col1_decimal");
        obj.setColType("decimal");
        ColumnStatisticsData data = new ColumnStatisticsData();
        DecimalColumnStatsData dcsd = new DecimalColumnStatsData();
        dcsd.setHighValue(StatObjectConverter.createThriftDecimal("" + (1000 + i)));
        dcsd.setLowValue(StatObjectConverter.createThriftDecimal("" + (-1000 - i)));
        dcsd.setNumNulls(i);
        dcsd.setNumDVs(10 * i + 1);
        dcsd.setBitVectors(bitVectors);
        data.setDecimalStats(dcsd);
        obj.setStatsData(data);
        cs.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(cs, partVal);
    }
    Checker statChecker = new Checker() {

        @Override
        public void checkStats(AggrStats aggrStats) throws Exception {
            Assert.assertEquals(10, aggrStats.getPartsFound());
            Assert.assertEquals(1, aggrStats.getColStatsSize());
            ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
            Assert.assertEquals("col1_decimal", cso.getColName());
            Assert.assertEquals("decimal", cso.getColType());
            DecimalColumnStatsData lcsd = cso.getStatsData().getDecimalStats();
            Assert.assertEquals(1009, HBaseUtils.getDoubleValue(lcsd.getHighValue()), 0.01);
            Assert.assertEquals(-1009, HBaseUtils.getDoubleValue(lcsd.getLowValue()), 0.01);
            Assert.assertEquals(45, lcsd.getNumNulls());
            Assert.assertEquals(3, lcsd.getNumDVs());
        }
    };
    List<String> partNames = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        partNames.add("ds=" + i);
    }
    AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, partNames, Arrays.asList("col1_decimal"));
    statChecker.checkStats(aggrStats);
}

Example 9 with ColumnStatistics

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

the class TestHBaseAggregateStatsExtrapolation method allPartitionsHaveBitVectorStatusLong.

@Test
public void allPartitionsHaveBitVectorStatusLong() throws Exception {
    String dbName = "default";
    String tableName = "snp";
    long now = System.currentTimeMillis();
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema("col1", "long", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, Collections.<String, String>emptyMap());
    List<FieldSchema> partCols = new ArrayList<>();
    partCols.add(new FieldSchema("ds", "string", ""));
    Table table = new Table(tableName, dbName, "me", (int) now, (int) now, 0, sd, partCols, Collections.<String, String>emptyMap(), null, null, null);
    store.createTable(table);
    List<List<String>> partVals = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        List<String> partVal = Arrays.asList("" + i);
        partVals.add(partVal);
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/default/hit/ds=" + partVal);
        Partition part = new Partition(partVal, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
        store.addPartition(part);
        ColumnStatistics cs = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
        desc.setLastAnalyzed(now);
        desc.setPartName("ds=" + partVal);
        cs.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName("col1");
        obj.setColType("long");
        ColumnStatisticsData data = new ColumnStatisticsData();
        LongColumnStatsData dcsd = new LongColumnStatsData();
        dcsd.setHighValue(1000 + i);
        dcsd.setLowValue(-1000 - i);
        dcsd.setNumNulls(i);
        dcsd.setNumDVs(10 * i + 1);
        dcsd.setBitVectors(bitVectors);
        data.setLongStats(dcsd);
        obj.setStatsData(data);
        cs.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(cs, partVal);
    }
    Checker statChecker = new Checker() {

        @Override
        public void checkStats(AggrStats aggrStats) throws Exception {
            Assert.assertEquals(10, aggrStats.getPartsFound());
            Assert.assertEquals(1, aggrStats.getColStatsSize());
            ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
            Assert.assertEquals("col1", cso.getColName());
            Assert.assertEquals("long", cso.getColType());
            LongColumnStatsData lcsd = cso.getStatsData().getLongStats();
            Assert.assertEquals(1009, lcsd.getHighValue(), 0.01);
            Assert.assertEquals(-1009, lcsd.getLowValue(), 0.01);
            Assert.assertEquals(45, lcsd.getNumNulls());
            Assert.assertEquals(3, lcsd.getNumDVs());
        }
    };
    List<String> partNames = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        partNames.add("ds=" + i);
    }
    AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, partNames, Arrays.asList("col1"));
    statChecker.checkStats(aggrStats);
}

Example 10 with ColumnStatistics

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

the class TestHBaseAggregateStatsExtrapolation method TwoEndsAndMiddleOfPartitionsHaveBitVectorStatusDouble.

@Test
public void TwoEndsAndMiddleOfPartitionsHaveBitVectorStatusDouble() throws Exception {
    String dbName = "default";
    String tableName = "snp";
    long now = System.currentTimeMillis();
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema("col5_double", "double", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, Collections.<String, String>emptyMap());
    List<FieldSchema> partCols = new ArrayList<>();
    partCols.add(new FieldSchema("ds", "string", ""));
    Table table = new Table(tableName, dbName, "me", (int) now, (int) now, 0, sd, partCols, Collections.<String, String>emptyMap(), null, null, null);
    store.createTable(table);
    List<List<String>> partVals = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        List<String> partVal = Arrays.asList("" + i);
        partVals.add(partVal);
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/default/hit/ds=" + partVal);
        Partition part = new Partition(partVal, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
        store.addPartition(part);
        if (i == 0 || i == 2 || i == 3 || i == 5 || i == 6 || i == 8) {
            ColumnStatistics cs = new ColumnStatistics();
            ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
            desc.setLastAnalyzed(now);
            desc.setPartName("ds=" + partVal);
            cs.setStatsDesc(desc);
            ColumnStatisticsObj obj = new ColumnStatisticsObj();
            obj.setColName("col5_double");
            obj.setColType("double");
            ColumnStatisticsData data = new ColumnStatisticsData();
            DoubleColumnStatsData dcsd = new DoubleColumnStatsData();
            dcsd.setHighValue(1000 + i);
            dcsd.setLowValue(-1000 - i);
            dcsd.setNumNulls(i);
            dcsd.setNumDVs(10 * i);
            dcsd.setBitVectors(bitVectors);
            data.setDoubleStats(dcsd);
            obj.setStatsData(data);
            cs.addToStatsObj(obj);
            store.updatePartitionColumnStatistics(cs, partVal);
        }
    }
    Checker statChecker = new Checker() {

        @Override
        public void checkStats(AggrStats aggrStats) throws Exception {
            Assert.assertEquals(6, aggrStats.getPartsFound());
            Assert.assertEquals(1, aggrStats.getColStatsSize());
            ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
            Assert.assertEquals("col5_double", cso.getColName());
            Assert.assertEquals("double", cso.getColType());
            DoubleColumnStatsData lcsd = cso.getStatsData().getDoubleStats();
            Assert.assertEquals(1010, lcsd.getHighValue(), 0.01);
            Assert.assertEquals(-1010, lcsd.getLowValue(), 0.01);
            Assert.assertEquals(40, lcsd.getNumNulls());
            Assert.assertEquals(3, lcsd.getNumDVs());
        }
    };
    List<String> partNames = new ArrayList<>();
    for (int i = 0; i < 10; i++) {
        partNames.add("ds=" + i);
    }
    AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, partNames, Arrays.asList("col5_double"));
    statChecker.checkStats(aggrStats);
}