Search in sources :

Example 16 with AggrStats

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

the class TestHBaseAggrStatsCacheIntegration method someWithStats.

@Test
public void someWithStats() throws Exception {
    String dbName = "default";
    String tableName = "psws";
    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", "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);
    boolean first = true;
    for (List<String> partVals : Arrays.asList(partVals1, partVals2)) {
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/default/psws/ds=" + partVals.get(0));
        Partition part = new Partition(partVals, dbName, tableName, (int) now, (int) now, psd, Collections.<String, String>emptyMap());
        store.addPartition(part);
        if (first) {
            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("long");
            ColumnStatisticsData data = new ColumnStatisticsData();
            LongColumnStatsData lcsd = new LongColumnStatsData();
            lcsd.setHighValue(192L);
            lcsd.setLowValue(-20L);
            lcsd.setNumNulls(30);
            lcsd.setNumDVs(32);
            data.setLongStats(lcsd);
            obj.setStatsData(data);
            cs.addToStatsObj(obj);
            store.updatePartitionColumnStatistics(cs, partVals);
            first = false;
        }
    }
    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("long", cso.getColType());
            LongColumnStatsData lcsd = cso.getStatsData().getLongStats();
            Assert.assertEquals(192L, lcsd.getHighValue());
            Assert.assertEquals(-20L, lcsd.getLowValue());
            Assert.assertEquals(30, lcsd.getNumNulls());
            Assert.assertEquals(32, lcsd.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());
    store.backdoor().getStatsCache().flushMemory();
    // Call again, this time it should come from hbase
    aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=today", "ds=yesterday"), Arrays.asList("col1"));
    statChecker.checkStats(aggrStats);
    Assert.assertEquals(1, store.backdoor().getStatsCache().hbaseHits.getCnt());
    Assert.assertEquals(3, store.backdoor().getStatsCache().totalGets.getCnt());
    Assert.assertEquals(1, store.backdoor().getStatsCache().misses.getCnt());
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) 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) LongColumnStatsData(org.apache.hadoop.hive.metastore.api.LongColumnStatsData) 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 17 with AggrStats

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

the class TestHBaseAggrStatsCacheIntegration method invalidation.

@Test
public void invalidation() throws Exception {
    try {
        String dbName = "default";
        String tableName = "invalidation";
        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);
        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());
            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());
        // Now call a different combination to get it in memory too
        aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=tomorrow", "ds=today"), Arrays.asList("col1"));
        statChecker.checkStats(aggrStats);
        Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
        Assert.assertEquals(3, store.backdoor().getStatsCache().totalGets.getCnt());
        Assert.assertEquals(2, store.backdoor().getStatsCache().misses.getCnt());
        aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=tomorrow", "ds=today"), Arrays.asList("col1"));
        statChecker.checkStats(aggrStats);
        Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
        Assert.assertEquals(4, 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();
        aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=tomorrow", "ds=today"), Arrays.asList("col1"));
        statChecker.checkStats(aggrStats);
        Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
        Assert.assertEquals(5, store.backdoor().getStatsCache().totalGets.getCnt());
        Assert.assertEquals(2, store.backdoor().getStatsCache().misses.getCnt());
        // Update statistics for 'tomorrow'
        ColumnStatistics cs = new ColumnStatistics();
        ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, dbName, tableName);
        desc.setLastAnalyzed(now);
        desc.setPartName("ds=" + partVals3.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(100);
        bcsd.setNumTrues(200);
        bcsd.setNumNulls(300);
        data.setBooleanStats(bcsd);
        obj.setStatsData(data);
        cs.addToStatsObj(obj);
        Checker afterUpdate = 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(110, bcsd.getNumFalses());
                Assert.assertEquals(220, bcsd.getNumTrues());
                Assert.assertEquals(330, bcsd.getNumNulls());
            }
        };
        store.updatePartitionColumnStatistics(cs, partVals3);
        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"));
        afterUpdate.checkStats(aggrStats);
        // Check that we missed, which means this aggregate was dropped from the cache.
        Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
        Assert.assertEquals(6, 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"));
        statChecker.checkStats(aggrStats);
        Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
        Assert.assertEquals(7, store.backdoor().getStatsCache().totalGets.getCnt());
        Assert.assertEquals(3, store.backdoor().getStatsCache().misses.getCnt());
        // Drop 'yesterday', so our first aggregate should be dumped from memory and hbase
        store.dropPartition(dbName, tableName, partVals2);
        store.backdoor().getStatsCache().wakeInvalidator();
        aggrStats = store.get_aggr_stats_for(dbName, tableName, Arrays.asList("ds=yesterday", "ds=today"), Arrays.asList("col1"));
        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("boolean", cso.getColType());
                BooleanColumnStatsData bcsd = cso.getStatsData().getBooleanStats();
                Assert.assertEquals(10, bcsd.getNumFalses());
                Assert.assertEquals(20, bcsd.getNumTrues());
                Assert.assertEquals(30, bcsd.getNumNulls());
            }
        }.checkStats(aggrStats);
        // Check that we missed, which means this aggregate was dropped from the cache.
        Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
        Assert.assertEquals(8, store.backdoor().getStatsCache().totalGets.getCnt());
        Assert.assertEquals(4, 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=tomorrow", "ds=today"), Arrays.asList("col1"));
        afterUpdate.checkStats(aggrStats);
        Assert.assertEquals(0, store.backdoor().getStatsCache().hbaseHits.getCnt());
        Assert.assertEquals(9, store.backdoor().getStatsCache().totalGets.getCnt());
        Assert.assertEquals(4, 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) ExpectedException(org.junit.rules.ExpectedException) IOException(java.io.IOException) 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 18 with AggrStats

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

the class TestHBaseAggrStatsCacheIntegration method altersInvalidation.

@Test
public void altersInvalidation() throws Exception {
    try {
        String dbName = "default";
        String tableName = "asi";
        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[] newParts = new Partition[2];
        newParts[0] = new Partition(partitions[0]);
        newParts[0].setLastAccessTime((int) System.currentTimeMillis());
        newParts[1] = new Partition(partitions[2]);
        newParts[1].setLastAccessTime((int) System.currentTimeMillis());
        store.alterPartitions(dbName, tableName, Arrays.asList(partVals1, partVals3), Arrays.asList(newParts));
        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 got dropped too
        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(4, 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 19 with AggrStats

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

the class TestHBaseAggregateStatsNDVUniformDist method TwoEndsOfPartitionsHaveBitVectorStatus.

@Test
public void TwoEndsOfPartitionsHaveBitVectorStatus() throws Exception {
    String dbName = "default";
    String tableName = "snp";
    long now = System.currentTimeMillis();
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema("col3", "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);
        if (i < 2 || i > 7) {
            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("col3");
            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[i / 5]);
            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(4, aggrStats.getPartsFound());
            Assert.assertEquals(1, aggrStats.getColStatsSize());
            ColumnStatisticsObj cso = aggrStats.getColStats().get(0);
            Assert.assertEquals("col3", cso.getColName());
            Assert.assertEquals("long", cso.getColType());
            LongColumnStatsData lcsd = cso.getStatsData().getLongStats();
            Assert.assertEquals(1010, lcsd.getHighValue(), 0.01);
            Assert.assertEquals(-1010, lcsd.getLowValue(), 0.01);
            Assert.assertEquals(45, lcsd.getNumNulls());
            Assert.assertEquals(12, 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("col3"));
    statChecker.checkStats(aggrStats);
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) 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) LongColumnStatsData(org.apache.hadoop.hive.metastore.api.LongColumnStatsData) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) ArrayList(java.util.ArrayList) List(java.util.List) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Example 20 with AggrStats

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

the class TestHBaseAggregateStatsNDVUniformDist method noPartitionsHaveBitVectorStatus.

@Test
public void noPartitionsHaveBitVectorStatus() throws Exception {
    String dbName = "default";
    String tableName = "snp";
    long now = System.currentTimeMillis();
    List<FieldSchema> cols = new ArrayList<>();
    cols.add(new FieldSchema("col2", "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("col2");
        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);
        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("col2", 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(91, 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("col2"));
    statChecker.checkStats(aggrStats);
}
Also used : ColumnStatistics(org.apache.hadoop.hive.metastore.api.ColumnStatistics) 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) LongColumnStatsData(org.apache.hadoop.hive.metastore.api.LongColumnStatsData) ColumnStatisticsObj(org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj) ColumnStatisticsDesc(org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc) ArrayList(java.util.ArrayList) List(java.util.List) ColumnStatisticsData(org.apache.hadoop.hive.metastore.api.ColumnStatisticsData) Test(org.junit.Test)

Aggregations

AggrStats (org.apache.hadoop.hive.metastore.api.AggrStats)32 ArrayList (java.util.ArrayList)31 ColumnStatisticsObj (org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj)27 FieldSchema (org.apache.hadoop.hive.metastore.api.FieldSchema)27 SerDeInfo (org.apache.hadoop.hive.metastore.api.SerDeInfo)26 StorageDescriptor (org.apache.hadoop.hive.metastore.api.StorageDescriptor)26 Table (org.apache.hadoop.hive.metastore.api.Table)26 Test (org.junit.Test)26 ColumnStatisticsData (org.apache.hadoop.hive.metastore.api.ColumnStatisticsData)25 Partition (org.apache.hadoop.hive.metastore.api.Partition)25 ColumnStatistics (org.apache.hadoop.hive.metastore.api.ColumnStatistics)24 ColumnStatisticsDesc (org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc)24 List (java.util.List)19 LongColumnStatsData (org.apache.hadoop.hive.metastore.api.LongColumnStatsData)11 BooleanColumnStatsData (org.apache.hadoop.hive.metastore.api.BooleanColumnStatsData)5 DoubleColumnStatsData (org.apache.hadoop.hive.metastore.api.DoubleColumnStatsData)5 IOException (java.io.IOException)2 DecimalColumnStatsData (org.apache.hadoop.hive.metastore.api.DecimalColumnStatsData)2 StringColumnStatsData (org.apache.hadoop.hive.metastore.api.StringColumnStatsData)2 Partition (org.apache.hadoop.hive.ql.metadata.Partition)2