Search in sources :

Example 16 with ColumnStatisticsData

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

the class TestHBaseStoreBitVector method stringPartitionStatistics.

@Test
public void stringPartitionStatistics() throws Exception {
    createMockTableAndPartition(STRING_TYPE, STRING_VAL);
    // Add partition stats for: STRING_COL and partition: {PART_KEY, STRING_VAL} to DB
    // Because of the way our mock implementation works we actually need to not create the table
    // before we set statistics on it.
    ColumnStatistics stats = new ColumnStatistics();
    // Get a default ColumnStatisticsDesc for partition level stats
    ColumnStatisticsDesc desc = getMockPartColStatsDesc(PART_KEY, STRING_VAL);
    stats.setStatsDesc(desc);
    // Get one of the pre-created ColumnStatisticsObj
    ColumnStatisticsObj obj = stringColStatsObjs.get(0);
    StringColumnStatsData stringData = obj.getStatsData().getStringStats();
    // Add to DB
    stats.addToStatsObj(obj);
    List<String> parVals = new ArrayList<String>();
    parVals.add(STRING_VAL);
    store.updatePartitionColumnStatistics(stats, parVals);
    // Get from DB
    List<String> partNames = new ArrayList<String>();
    partNames.add(desc.getPartName());
    List<String> colNames = new ArrayList<String>();
    colNames.add(obj.getColName());
    List<ColumnStatistics> statsFromDB = store.getPartitionColumnStatistics(DB, TBL, partNames, colNames);
    // Compare ColumnStatisticsDesc
    Assert.assertEquals(1, statsFromDB.size());
    Assert.assertEquals(desc.getLastAnalyzed(), statsFromDB.get(0).getStatsDesc().getLastAnalyzed());
    Assert.assertEquals(DB, statsFromDB.get(0).getStatsDesc().getDbName());
    Assert.assertEquals(TBL, statsFromDB.get(0).getStatsDesc().getTableName());
    Assert.assertFalse(statsFromDB.get(0).getStatsDesc().isIsTblLevel());
    // Compare ColumnStatisticsObj
    Assert.assertEquals(1, statsFromDB.get(0).getStatsObjSize());
    ColumnStatisticsObj objFromDB = statsFromDB.get(0).getStatsObj().get(0);
    ColumnStatisticsData dataFromDB = objFromDB.getStatsData();
    // Compare ColumnStatisticsData
    Assert.assertEquals(ColumnStatisticsData._Fields.STRING_STATS, dataFromDB.getSetField());
    // Compare StringColumnStatsData
    StringColumnStatsData stringDataFromDB = dataFromDB.getStringStats();
    Assert.assertEquals(stringData.getMaxColLen(), stringDataFromDB.getMaxColLen());
    Assert.assertEquals(stringData.getAvgColLen(), stringDataFromDB.getAvgColLen(), 0.01);
    Assert.assertEquals(stringData.getNumNulls(), stringDataFromDB.getNumNulls());
    Assert.assertEquals(stringData.getNumDVs(), stringDataFromDB.getNumDVs());
    Assert.assertEquals(stringData.getBitVectors(), stringDataFromDB.getBitVectors());
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) ArrayList(java.util.ArrayList) StringColumnStatsData(org.apache.hadoop.hive.metastore.api.StringColumnStatsData) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Example 17 with ColumnStatisticsData

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

the class TestHBaseStoreBitVector method decimalPartitionStatistics.

@Test
public void decimalPartitionStatistics() throws Exception {
    createMockTableAndPartition(DECIMAL_TYPE, DECIMAL_VAL);
    // Add partition stats for: DECIMAL_COL and partition: {PART_KEY, DECIMAL_VAL} to DB
    // Because of the way our mock implementation works we actually need to not create the table
    // before we set statistics on it.
    ColumnStatistics stats = new ColumnStatistics();
    // Get a default ColumnStatisticsDesc for partition level stats
    ColumnStatisticsDesc desc = getMockPartColStatsDesc(PART_KEY, DECIMAL_VAL);
    stats.setStatsDesc(desc);
    // Get one of the pre-created ColumnStatisticsObj
    ColumnStatisticsObj obj = decimalColStatsObjs.get(0);
    DecimalColumnStatsData decimalData = obj.getStatsData().getDecimalStats();
    // Add to DB
    stats.addToStatsObj(obj);
    List<String> parVals = new ArrayList<String>();
    parVals.add(DECIMAL_VAL);
    store.updatePartitionColumnStatistics(stats, parVals);
    // Get from DB
    List<String> partNames = new ArrayList<String>();
    partNames.add(desc.getPartName());
    List<String> colNames = new ArrayList<String>();
    colNames.add(obj.getColName());
    List<ColumnStatistics> statsFromDB = store.getPartitionColumnStatistics(DB, TBL, partNames, colNames);
    // Compare ColumnStatisticsDesc
    Assert.assertEquals(1, statsFromDB.size());
    Assert.assertEquals(desc.getLastAnalyzed(), statsFromDB.get(0).getStatsDesc().getLastAnalyzed());
    Assert.assertEquals(DB, statsFromDB.get(0).getStatsDesc().getDbName());
    Assert.assertEquals(TBL, statsFromDB.get(0).getStatsDesc().getTableName());
    Assert.assertFalse(statsFromDB.get(0).getStatsDesc().isIsTblLevel());
    // Compare ColumnStatisticsObj
    Assert.assertEquals(1, statsFromDB.get(0).getStatsObjSize());
    ColumnStatisticsObj objFromDB = statsFromDB.get(0).getStatsObj().get(0);
    ColumnStatisticsData dataFromDB = objFromDB.getStatsData();
    // Compare ColumnStatisticsData
    Assert.assertEquals(ColumnStatisticsData._Fields.DECIMAL_STATS, dataFromDB.getSetField());
    // Compare DecimalColumnStatsData
    DecimalColumnStatsData decimalDataFromDB = dataFromDB.getDecimalStats();
    Assert.assertEquals(decimalData.getHighValue(), decimalDataFromDB.getHighValue());
    Assert.assertEquals(decimalData.getLowValue(), decimalDataFromDB.getLowValue());
    Assert.assertEquals(decimalData.getNumNulls(), decimalDataFromDB.getNumNulls());
    Assert.assertEquals(decimalData.getNumDVs(), decimalDataFromDB.getNumDVs());
    Assert.assertEquals(decimalData.getBitVectors(), decimalDataFromDB.getBitVectors());
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) DecimalColumnStatsData(org.apache.hadoop.hive.metastore.api.DecimalColumnStatsData) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) ArrayList(java.util.ArrayList) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Example 18 with ColumnStatisticsData

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

the class TestHBaseStoreIntegration method partitionStatistics.

@Test
public void partitionStatistics() throws Exception {
    long now = System.currentTimeMillis();
    String dbname = "default";
    String tableName = "statspart";
    String[] partNames = { "ds=today", "ds=yesterday" };
    String[] partVals = { "today", "yesterday" };
    String boolcol = "boolcol";
    String longcol = "longcol";
    String doublecol = "doublecol";
    String stringcol = "stringcol";
    String binarycol = "bincol";
    String decimalcol = "deccol";
    long trues = 37;
    long falses = 12;
    long booleanNulls = 2;
    long strMaxLen = 1234;
    double strAvgLen = 32.3;
    long strNulls = 987;
    long strDVs = 906;
    List<FieldSchema> cols = new ArrayList<FieldSchema>();
    cols.add(new FieldSchema(boolcol, "boolean", "nocomment"));
    cols.add(new FieldSchema(longcol, "long", "nocomment"));
    cols.add(new FieldSchema(doublecol, "double", "nocomment"));
    cols.add(new FieldSchema(stringcol, "varchar(32)", "nocomment"));
    cols.add(new FieldSchema(binarycol, "binary", "nocomment"));
    cols.add(new FieldSchema(decimalcol, "decimal(5, 3)", "nocomment"));
    SerDeInfo serde = new SerDeInfo("serde", "seriallib", null);
    StorageDescriptor sd = new StorageDescriptor(cols, "file:/tmp", "input", "output", false, 0, serde, null, null, emptyParameters);
    List<FieldSchema> partCols = new ArrayList<FieldSchema>();
    partCols.add(new FieldSchema("ds", "string", ""));
    Table table = new Table(tableName, dbname, "me", (int) now / 1000, (int) now / 1000, 0, sd, partCols, emptyParameters, null, null, null);
    store.createTable(table);
    for (String partVal : partVals) {
        Partition part = new Partition(Arrays.asList(partVal), dbname, tableName, (int) now / 1000, (int) now / 1000, sd, emptyParameters);
        store.addPartition(part);
    }
    for (int i = 0; i < partNames.length; i++) {
        ColumnStatistics stats = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc();
        desc.setLastAnalyzed(now);
        desc.setDbName(dbname);
        desc.setTableName(tableName);
        desc.setIsTblLevel(false);
        desc.setPartName(partNames[i]);
        stats.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName(boolcol);
        obj.setColType("boolean");
        ColumnStatisticsData data = new ColumnStatisticsData();
        BooleanColumnStatsData boolData = new BooleanColumnStatsData();
        boolData.setNumTrues(trues);
        boolData.setNumFalses(falses);
        boolData.setNumNulls(booleanNulls);
        data.setBooleanStats(boolData);
        obj.setStatsData(data);
        stats.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(stats, Arrays.asList(partVals[i]));
    }
    List<ColumnStatistics> statsList = store.getPartitionColumnStatistics(dbname, tableName, Arrays.asList(partNames), Arrays.asList(boolcol));
    Assert.assertEquals(2, statsList.size());
    for (int i = 0; i < partNames.length; i++) {
        Assert.assertEquals(1, statsList.get(i).getStatsObjSize());
    }
    for (int i = 0; i < partNames.length; i++) {
        ColumnStatistics stats = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc();
        desc.setLastAnalyzed(now);
        desc.setDbName(dbname);
        desc.setTableName(tableName);
        desc.setIsTblLevel(false);
        desc.setPartName(partNames[i]);
        stats.setStatsDesc(desc);
        ColumnStatisticsObj obj = new ColumnStatisticsObj();
        obj.setColName(stringcol);
        obj.setColType("string");
        ColumnStatisticsData data = new ColumnStatisticsData();
        StringColumnStatsData strData = new StringColumnStatsData();
        strData.setMaxColLen(strMaxLen);
        strData.setAvgColLen(strAvgLen);
        strData.setNumNulls(strNulls);
        strData.setNumDVs(strDVs);
        data.setStringStats(strData);
        obj.setStatsData(data);
        stats.addToStatsObj(obj);
        store.updatePartitionColumnStatistics(stats, Arrays.asList(partVals[i]));
    }
    // Make sure when we ask for one we only get one
    statsList = store.getPartitionColumnStatistics(dbname, tableName, Arrays.asList(partNames), Arrays.asList(boolcol));
    Assert.assertEquals(2, statsList.size());
    for (int i = 0; i < partNames.length; i++) {
        Assert.assertEquals(1, statsList.get(i).getStatsObjSize());
    }
    statsList = store.getPartitionColumnStatistics(dbname, tableName, Arrays.asList(partNames), Arrays.asList(boolcol, stringcol));
    Assert.assertEquals(2, statsList.size());
    for (int i = 0; i < partNames.length; i++) {
        Assert.assertEquals(2, statsList.get(i).getStatsObjSize());
        // Just check one piece of the data, I don't need to check it all again
        Assert.assertEquals(booleanNulls, statsList.get(i).getStatsObj().get(0).getStatsData().getBooleanStats().getNumNulls());
        Assert.assertEquals(strDVs, statsList.get(i).getStatsObj().get(1).getStatsData().getStringStats().getNumDVs());
    }
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) BooleanColumnStatsData(org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData) Partition(org.apache.hadoop.hive.metastore.api.Partition) Table(org.apache.hadoop.hive.metastore.api.Table) FieldSchema(org.apache.hadoop.hive.metastore.api.FieldSchema) SerDeInfo(org.apache.hadoop.hive.metastore.api.SerDeInfo) ArrayList(java.util.ArrayList) StorageDescriptor(org.apache.hadoop.hive.metastore.api.StorageDescriptor) StringColumnStatsData(org.apache.hadoop.hive.metastore.api.StringColumnStatsData) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Example 19 with ColumnStatisticsData

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

the class TestHBaseAggrStatsCacheIntegration method alterInvalidation.

@Test
public void alterInvalidation() throws Exception {
    try {
        String dbName = "default";
        String tableName = "ai";
        List<String> partVals1 = Arrays.asList("today");
        List<String> partVals2 = Arrays.asList("yesterday");
        List<String> partVals3 = Arrays.asList("tomorrow");
        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);
        Partition[] partitions = new Partition[3];
        int partnum = 0;
        for (List<String> partVals : Arrays.asList(partVals1, partVals2, partVals3)) {
            StorageDescriptor psd = new StorageDescriptor(sd);
            psd.setLocation("file:/tmp/default/invalidation/ds=" + partVals.get(0));
            Partition part = new Partition(partVals, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
            partitions[partnum++] = part;
            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);
        }
        AggrStats aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=tomorrow"), Arrays.asList("col1"));
        aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=yesterday"), Arrays.asList("col1"));
        // Check that we had to build it from the stats
        Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
        Assert.assertEquals(2, store.backdoor().getStatsCache().totalGets.getCnt());
        Assert.assertEquals(2, store.backdoor().getStatsCache().misses.getCnt());
        // wake the invalidator and check again to make sure it isn't too aggressive about
        // removing our stuff.
        store.backdoor().getStatsCache().wakeInvalidator();
        Partition newPart = new Partition(partitions[2]);
        newPart.setLastAccessTime((int) System.currentTimeMillis());
        store.alterPartition(dbName, tableName, partVals3, newPart);
        store.backdoor().getStatsCache().setRunInvalidatorEvery(100);
        store.backdoor().getStatsCache().wakeInvalidator();
        aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=tomorrow", "ds=today"), Arrays.asList("col1"));
        // Check that we missed, which means this aggregate was dropped from the cache.
        Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
        Assert.assertEquals(3, store.backdoor().getStatsCache().totalGets.getCnt());
        Assert.assertEquals(3, store.backdoor().getStatsCache().misses.getCnt());
        // Check that our other aggregate is still in the cache.
        aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=yesterday", "ds=today"), Arrays.asList("col1"));
        Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
        Assert.assertEquals(4, store.backdoor().getStatsCache().totalGets.getCnt());
        Assert.assertEquals(3, store.backdoor().getStatsCache().misses.getCnt());
    } finally {
        store.backdoor().getStatsCache().setRunInvalidatorEvery(5000);
        store.backdoor().getStatsCache().setMaxTimeInCache(500000);
        store.backdoor().getStatsCache().wakeInvalidator();
    }
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) BooleanColumnStatsData(org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData) Partition(org.apache.hadoop.hive.metastore.api.Partition) Table(org.apache.hadoop.hive.metastore.api.Table) AggrStats(org.apache.hadoop.hive.metastore.api.AggrStats) FieldSchema(org.apache.hadoop.hive.metastore.api.FieldSchema) SerDeInfo(org.apache.hadoop.hive.metastore.api.SerDeInfo) ArrayList(java.util.ArrayList) StorageDescriptor(org.apache.hadoop.hive.metastore.api.StorageDescriptor) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Example 20 with ColumnStatisticsData

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

the class TestHBaseStore method mockLongStats.

private static ColumnStatisticsObj mockLongStats(int i) {
    long high = 120938479124L + 100 * i;
    long low = -12341243213412124L - 50 * i;
    long nulls = 23 + i;
    long dVs = 213L + 10 * i;
    ColumnStatisticsObj colStatsObj = new ColumnStatisticsObj();
    colStatsObj.setColName(LONG_COL);
    colStatsObj.setColType(LONG_TYPE);
    ColumnStatisticsData data = new ColumnStatisticsData();
    LongColumnStatsData longData = new LongColumnStatsData();
    longData.setHighValue(high);
    longData.setLowValue(low);
    longData.setNumNulls(nulls);
    longData.setNumDVs(dVs);
    data.setLongStats(longData);
    colStatsObj.setStatsData(data);
    return colStatsObj;
}
Also used : ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) LongColumnStatsData(org.apache.hadoop.hive.metastore.api.LongColumnStatsData) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData)

Aggregations

ColumnStatisticsData (org.apache.hadoop.hive.metastore.api.ColumnStatisticsData)127 ColumnStatisticsObj (org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj)104 ColumnStatistics (org.apache.hadoop.hive.metastore.api.ColumnStatistics)61 Test (org.junit.Test)57 ColumnStatisticsDesc (org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc)54 ArrayList (java.util.ArrayList)49 LongColumnStatsData (org.apache.hadoop.hive.metastore.api.LongColumnStatsData)39 BooleanColumnStatsData (org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData)34 FieldSchema (org.apache.hadoop.hive.metastore.api.FieldSchema)34 Table (org.apache.hadoop.hive.metastore.api.Table)31 DoubleColumnStatsData (org.apache.hadoop.hive.metastore.api.DoubleColumnStatsData)30 StorageDescriptor (org.apache.hadoop.hive.metastore.api.StorageDescriptor)29 Partition (org.apache.hadoop.hive.metastore.api.Partition)28 SerDeInfo (org.apache.hadoop.hive.metastore.api.SerDeInfo)28 StringColumnStatsData (org.apache.hadoop.hive.metastore.api.StringColumnStatsData)28 AggrStats (org.apache.hadoop.hive.metastore.api.AggrStats)26 BinaryColumnStatsData (org.apache.hadoop.hive.metastore.api.BinaryColumnStatsData)26 DecimalColumnStatsData (org.apache.hadoop.hive.metastore.api.DecimalColumnStatsData)26 HashMap (java.util.HashMap)20 List (java.util.List)19