Search in sources :

Example 36 with Partition

use of org.apache.hadoop.hive.metastore.api.Partition 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 37 with Partition

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

the class TestReplicationScenarios method testBasicWithCM.

@Test
public void testBasicWithCM() throws Exception {
    String testName = "basic_with_cm";
    LOG.info("Testing " + testName);
    String dbName = testName + "_" + tid;
    run("CREATE DATABASE " + dbName);
    run("CREATE TABLE " + dbName + ".unptned(a string) STORED AS TEXTFILE");
    run("CREATE TABLE " + dbName + ".ptned(a string) partitioned by (b int) STORED AS TEXTFILE");
    run("CREATE TABLE " + dbName + ".unptned_empty(a string) STORED AS TEXTFILE");
    run("CREATE TABLE " + dbName + ".ptned_empty(a string) partitioned by (b int) STORED AS TEXTFILE");
    String[] unptn_data = new String[] { "eleven", "twelve" };
    String[] ptn_data_1 = new String[] { "thirteen", "fourteen", "fifteen" };
    String[] ptn_data_2 = new String[] { "fifteen", "sixteen", "seventeen" };
    String[] ptn_data_2_later = new String[] { "eighteen", "nineteen", "twenty" };
    String[] empty = new String[] {};
    String unptn_locn = new Path(TEST_PATH, testName + "_unptn").toUri().getPath();
    String ptn_locn_1 = new Path(TEST_PATH, testName + "_ptn1").toUri().getPath();
    String ptn_locn_2 = new Path(TEST_PATH, testName + "_ptn2").toUri().getPath();
    String ptn_locn_2_later = new Path(TEST_PATH, testName + "_ptn2_later").toUri().getPath();
    createTestDataFile(unptn_locn, unptn_data);
    createTestDataFile(ptn_locn_1, ptn_data_1);
    createTestDataFile(ptn_locn_2, ptn_data_2);
    createTestDataFile(ptn_locn_2_later, ptn_data_2_later);
    run("LOAD DATA LOCAL INPATH '" + unptn_locn + "' OVERWRITE INTO TABLE " + dbName + ".unptned");
    run("SELECT * from " + dbName + ".unptned");
    verifyResults(unptn_data);
    run("LOAD DATA LOCAL INPATH '" + ptn_locn_1 + "' OVERWRITE INTO TABLE " + dbName + ".ptned PARTITION(b=1)");
    run("SELECT a from " + dbName + ".ptned WHERE b=1");
    verifyResults(ptn_data_1);
    run("LOAD DATA LOCAL INPATH '" + ptn_locn_2 + "' OVERWRITE INTO TABLE " + dbName + ".ptned PARTITION(b=2)");
    run("SELECT a from " + dbName + ".ptned WHERE b=2");
    verifyResults(ptn_data_2);
    run("SELECT a from " + dbName + ".ptned_empty");
    verifyResults(empty);
    run("SELECT * from " + dbName + ".unptned_empty");
    verifyResults(empty);
    advanceDumpDir();
    run("REPL DUMP " + dbName);
    String replDumpLocn = getResult(0, 0);
    String replDumpId = getResult(0, 1, true);
    // Table dropped after "repl dump"
    run("DROP TABLE " + dbName + ".unptned");
    // Partition droppped after "repl dump"
    run("ALTER TABLE " + dbName + ".ptned " + "DROP PARTITION(b=1)");
    // File changed after "repl dump"
    Partition p = metaStoreClient.getPartition(dbName, "ptned", "b=2");
    Path loc = new Path(p.getSd().getLocation());
    FileSystem fs = loc.getFileSystem(hconf);
    Path file = fs.listStatus(loc)[0].getPath();
    fs.delete(file, false);
    fs.copyFromLocalFile(new Path(ptn_locn_2_later), file);
    run("EXPLAIN REPL LOAD " + dbName + "_dupe FROM '" + replDumpLocn + "'");
    printOutput();
    run("REPL LOAD " + dbName + "_dupe FROM '" + replDumpLocn + "'");
    run("REPL STATUS " + dbName + "_dupe");
    verifyResults(new String[] { replDumpId });
    run("SELECT * from " + dbName + "_dupe.unptned");
    verifyResults(unptn_data);
    run("SELECT a from " + dbName + "_dupe.ptned WHERE b=1");
    verifyResults(ptn_data_1);
    // Since partition(b=2) changed manually, Hive cannot find
    // it in original location and cmroot, thus empty
    run("SELECT a from " + dbName + "_dupe.ptned WHERE b=2");
    verifyResults(empty);
    run("SELECT a from " + dbName + ".ptned_empty");
    verifyResults(empty);
    run("SELECT * from " + dbName + ".unptned_empty");
    verifyResults(empty);
}
Also used : Path(org.apache.hadoop.fs.Path) Partition(org.apache.hadoop.hive.metastore.api.Partition) FileSystem(org.apache.hadoop.fs.FileSystem) Test(org.junit.Test)

Example 38 with Partition

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

the class TestHBaseStoreIntegration method alterPartitions.

@Test
public void alterPartitions() throws Exception {
    String dbName = "default";
    String tableName = "alterParts";
    int startTime = (int) (System.currentTimeMillis() / 1000);
    List<FieldSchema> cols = new ArrayList<FieldSchema>();
    cols.add(new FieldSchema("col1", "int", "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("pc", "string", ""));
    Table table = new Table(tableName, dbName, "me", startTime, startTime, 0, sd, partCols, emptyParameters, null, null, null);
    store.createTable(table);
    List<String> partVals = Arrays.asList("alan", "bob", "carl", "doug", "ethan");
    List<Partition> partitions = new ArrayList<Partition>();
    List<List<String>> allVals = new ArrayList<List<String>>();
    for (String val : partVals) {
        List<String> vals = new ArrayList<String>();
        allVals.add(vals);
        vals.add(val);
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/pc=" + val);
        Partition part = new Partition(vals, dbName, tableName, startTime, startTime, psd, emptyParameters);
        partitions.add(part);
    }
    store.addPartitions(dbName, tableName, partitions);
    for (Partition p : partitions) p.setLastAccessTime(startTime + 10);
    store.alterPartitions(dbName, tableName, allVals, partitions);
    partitions = store.getPartitions(dbName, tableName, -1);
    for (Partition part : partitions) {
        Assert.assertEquals(startTime + 10, part.getLastAccessTime());
    }
}
Also used : 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) List(java.util.List) ArrayList(java.util.ArrayList) Test(org.junit.Test)

Example 39 with Partition

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

the class TestHBaseStoreIntegration method getPartitions.

@Test
public void getPartitions() throws Exception {
    String dbName = "default";
    String tableName = "manyParts";
    int startTime = (int) (System.currentTimeMillis() / 1000);
    List<FieldSchema> cols = new ArrayList<FieldSchema>();
    cols.add(new FieldSchema("col1", "int", "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("pc", "string", ""));
    Table table = new Table(tableName, dbName, "me", startTime, startTime, 0, sd, partCols, emptyParameters, null, null, null);
    store.createTable(table);
    List<String> partVals = Arrays.asList("alan", "bob", "carl", "doug", "ethan");
    for (String val : partVals) {
        List<String> vals = new ArrayList<String>();
        vals.add(val);
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/pc=" + val);
        Partition part = new Partition(vals, dbName, tableName, startTime, startTime, psd, emptyParameters);
        store.addPartition(part);
        Partition p = store.getPartition(dbName, tableName, vals);
        Assert.assertEquals("file:/tmp/pc=" + val, p.getSd().getLocation());
    }
    List<Partition> parts = store.getPartitions(dbName, tableName, -1);
    Assert.assertEquals(5, parts.size());
    String[] pv = new String[5];
    for (int i = 0; i < 5; i++) pv[i] = parts.get(i).getValues().get(0);
    Arrays.sort(pv);
    Assert.assertArrayEquals(pv, partVals.toArray(new String[5]));
}
Also used : 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) Test(org.junit.Test)

Example 40 with Partition

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

the class TestHBaseStoreIntegration method addPartitions.

@Test
public void addPartitions() throws Exception {
    String dbName = "default";
    String tableName = "addParts";
    int startTime = (int) (System.currentTimeMillis() / 1000);
    List<FieldSchema> cols = new ArrayList<FieldSchema>();
    cols.add(new FieldSchema("col1", "int", "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("pc", "string", ""));
    Table table = new Table(tableName, dbName, "me", startTime, startTime, 0, sd, partCols, emptyParameters, null, null, null);
    store.createTable(table);
    List<String> partVals = Arrays.asList("alan", "bob", "carl", "doug", "ethan");
    List<Partition> partitions = new ArrayList<Partition>();
    for (String val : partVals) {
        List<String> vals = new ArrayList<String>();
        vals.add(val);
        StorageDescriptor psd = new StorageDescriptor(sd);
        psd.setLocation("file:/tmp/pc=" + val);
        Partition part = new Partition(vals, dbName, tableName, startTime, startTime, psd, emptyParameters);
        partitions.add(part);
    }
    store.addPartitions(dbName, tableName, partitions);
    List<String> partNames = store.listPartitionNames(dbName, tableName, (short) -1);
    Assert.assertEquals(5, partNames.size());
    String[] names = partNames.toArray(new String[partNames.size()]);
    Arrays.sort(names);
    String[] canonicalNames = partVals.toArray(new String[partVals.size()]);
    for (int i = 0; i < canonicalNames.length; i++) canonicalNames[i] = "pc=" + canonicalNames[i];
    Assert.assertArrayEquals(canonicalNames, names);
}
Also used : 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) Test(org.junit.Test)

Aggregations

Partition (org.apache.hadoop.hive.metastore.api.Partition)466 Test (org.junit.Test)311 MetastoreCheckinTest (org.apache.hadoop.hive.metastore.annotation.MetastoreCheckinTest)213 Table (org.apache.hadoop.hive.metastore.api.Table)207 ArrayList (java.util.ArrayList)202 FieldSchema (org.apache.hadoop.hive.metastore.api.FieldSchema)97 StorageDescriptor (org.apache.hadoop.hive.metastore.api.StorageDescriptor)88 MetaException (org.apache.hadoop.hive.metastore.api.MetaException)72 SerDeInfo (org.apache.hadoop.hive.metastore.api.SerDeInfo)71 Path (org.apache.hadoop.fs.Path)56 List (java.util.List)55 NoSuchObjectException (org.apache.hadoop.hive.metastore.api.NoSuchObjectException)51 PartitionSpecProxy (org.apache.hadoop.hive.metastore.partition.spec.PartitionSpecProxy)41 Database (org.apache.hadoop.hive.metastore.api.Database)38 HashMap (java.util.HashMap)36 IOException (java.io.IOException)34 TException (org.apache.thrift.TException)34 ColumnStatistics (org.apache.hadoop.hive.metastore.api.ColumnStatistics)33 ColumnStatisticsObj (org.apache.hadoop.hive.metastore.api.ColumnStatisticsObj)33 ColumnStatisticsDesc (org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc)31