Search in sources :

Example 36 with IMetaStoreClient

use of org.apache.hadoop.hive.metastore.IMetaStoreClient in project hive by apache.

the class TestCompactor method testMinorCompactionForSplitUpdateWithInsertsAndDeletes.

@Test
public void testMinorCompactionForSplitUpdateWithInsertsAndDeletes() throws Exception {
    String agentInfo = "UT_" + Thread.currentThread().getName();
    String dbName = "default";
    String tblName = "cws";
    List<String> colNames = Arrays.asList("a", "b");
    String columnNamesProperty = "a,b";
    String columnTypesProperty = "int:string";
    executeStatementOnDriver("drop table if exists " + tblName, driver);
    executeStatementOnDriver("CREATE TABLE " + tblName + "(a INT, b STRING) " + // currently ACID requires table to be bucketed
    " CLUSTERED BY(a) INTO 1 BUCKETS" + " STORED AS ORC  TBLPROPERTIES ('transactional'='true'," + "'transactional_properties'='default')", driver);
    // Insert some data -> this will generate only insert deltas and no delete deltas: delta_3_3
    executeStatementOnDriver("INSERT INTO " + tblName + "(a,b) VALUES(1, 'foo')", driver);
    // Insert some data -> this will again generate only insert deltas and no delete deltas: delta_4_4
    executeStatementOnDriver("INSERT INTO " + tblName + "(a,b) VALUES(2, 'bar')", driver);
    // Delete some data -> this will generate only delete deltas and no insert deltas: delete_delta_5_5
    executeStatementOnDriver("DELETE FROM " + tblName + " WHERE a = 2", driver);
    // Now, compact -> Compaction produces a single range for both delta and delete delta
    // That is, both delta and delete_deltas would be compacted into delta_3_5 and delete_delta_3_5
    // even though there are only two delta_3_3, delta_4_4 and one delete_delta_5_5.
    TxnStore txnHandler = TxnUtils.getTxnStore(conf);
    txnHandler.compact(new CompactionRequest(dbName, tblName, CompactionType.MINOR));
    Worker t = new Worker();
    t.setThreadId((int) t.getId());
    t.setConf(conf);
    AtomicBoolean stop = new AtomicBoolean(true);
    AtomicBoolean looped = new AtomicBoolean();
    t.init(stop, looped);
    t.run();
    // Find the location of the table
    IMetaStoreClient msClient = new HiveMetaStoreClient(conf);
    Table table = msClient.getTable(dbName, tblName);
    FileSystem fs = FileSystem.get(conf);
    // Verify that we have got correct set of deltas.
    FileStatus[] stat = fs.listStatus(new Path(table.getSd().getLocation()), AcidUtils.deltaFileFilter);
    String[] deltas = new String[stat.length];
    Path minorCompactedDelta = null;
    for (int i = 0; i < deltas.length; i++) {
        deltas[i] = stat[i].getPath().getName();
        if (deltas[i].equals("delta_0000001_0000003")) {
            minorCompactedDelta = stat[i].getPath();
        }
    }
    Arrays.sort(deltas);
    String[] expectedDeltas = new String[] { "delta_0000001_0000001_0000", "delta_0000001_0000003", "delta_0000002_0000002_0000" };
    if (!Arrays.deepEquals(expectedDeltas, deltas)) {
        Assert.fail("Expected: " + Arrays.toString(expectedDeltas) + ", found: " + Arrays.toString(deltas));
    }
    checkExpectedTxnsPresent(null, new Path[] { minorCompactedDelta }, columnNamesProperty, columnTypesProperty, 0, 1L, 2L, 1);
    // Verify that we have got correct set of delete_deltas.
    FileStatus[] deleteDeltaStat = fs.listStatus(new Path(table.getSd().getLocation()), AcidUtils.deleteEventDeltaDirFilter);
    String[] deleteDeltas = new String[deleteDeltaStat.length];
    Path minorCompactedDeleteDelta = null;
    for (int i = 0; i < deleteDeltas.length; i++) {
        deleteDeltas[i] = deleteDeltaStat[i].getPath().getName();
        if (deleteDeltas[i].equals("delete_delta_0000001_0000003")) {
            minorCompactedDeleteDelta = deleteDeltaStat[i].getPath();
        }
    }
    Arrays.sort(deleteDeltas);
    String[] expectedDeleteDeltas = new String[] { "delete_delta_0000001_0000003", "delete_delta_0000003_0000003_0000" };
    if (!Arrays.deepEquals(expectedDeleteDeltas, deleteDeltas)) {
        Assert.fail("Expected: " + Arrays.toString(expectedDeleteDeltas) + ", found: " + Arrays.toString(deleteDeltas));
    }
    checkExpectedTxnsPresent(null, new Path[] { minorCompactedDeleteDelta }, columnNamesProperty, columnTypesProperty, 0, 2L, 2L, 1);
}
Also used : Path(org.apache.hadoop.fs.Path) HiveMetaStoreClient(org.apache.hadoop.hive.metastore.HiveMetaStoreClient) Table(org.apache.hadoop.hive.metastore.api.Table) FileStatus(org.apache.hadoop.fs.FileStatus) IMetaStoreClient(org.apache.hadoop.hive.metastore.IMetaStoreClient) HiveEndPoint(org.apache.hive.hcatalog.streaming.HiveEndPoint) AtomicBoolean(java.util.concurrent.atomic.AtomicBoolean) FileSystem(org.apache.hadoop.fs.FileSystem) TxnStore(org.apache.hadoop.hive.metastore.txn.TxnStore) CompactionRequest(org.apache.hadoop.hive.metastore.api.CompactionRequest) Test(org.junit.Test)

Example 37 with IMetaStoreClient

use of org.apache.hadoop.hive.metastore.IMetaStoreClient in project hive by apache.

the class TestCompactor method minorCompactAfterAbort.

@Test
public void minorCompactAfterAbort() throws Exception {
    String agentInfo = "UT_" + Thread.currentThread().getName();
    String dbName = "default";
    String tblName = "cws";
    List<String> colNames = Arrays.asList("a", "b");
    String columnNamesProperty = "a,b";
    String columnTypesProperty = "int:string";
    executeStatementOnDriver("drop table if exists " + tblName, driver);
    executeStatementOnDriver("CREATE TABLE " + tblName + "(a INT, b STRING) " + // currently ACID requires table to be bucketed
    " CLUSTERED BY(a) INTO 1 BUCKETS" + " STORED AS ORC  TBLPROPERTIES ('transactional'='true')", driver);
    HiveEndPoint endPt = new HiveEndPoint(null, dbName, tblName, null);
    DelimitedInputWriter writer = new DelimitedInputWriter(new String[] { "a", "b" }, ",", endPt);
    StreamingConnection connection = endPt.newConnection(false, "UT_" + Thread.currentThread().getName());
    try {
        // Write a couple of batches
        for (int i = 0; i < 2; i++) {
            writeBatch(connection, writer, false);
        }
        // Start a third batch, abort everything, don't properly close it
        TransactionBatch txnBatch = connection.fetchTransactionBatch(2, writer);
        txnBatch.beginNextTransaction();
        txnBatch.abort();
        txnBatch.beginNextTransaction();
        txnBatch.abort();
        // Now, compact
        TxnStore txnHandler = TxnUtils.getTxnStore(conf);
        txnHandler.compact(new CompactionRequest(dbName, tblName, CompactionType.MINOR));
        Worker t = new Worker();
        t.setThreadId((int) t.getId());
        t.setConf(conf);
        AtomicBoolean stop = new AtomicBoolean(true);
        AtomicBoolean looped = new AtomicBoolean();
        t.init(stop, looped);
        t.run();
        // Find the location of the table
        IMetaStoreClient msClient = new HiveMetaStoreClient(conf);
        Table table = msClient.getTable(dbName, tblName);
        FileSystem fs = FileSystem.get(conf);
        FileStatus[] stat = fs.listStatus(new Path(table.getSd().getLocation()), AcidUtils.deltaFileFilter);
        String[] names = new String[stat.length];
        Path resultDelta = null;
        for (int i = 0; i < names.length; i++) {
            names[i] = stat[i].getPath().getName();
            if (names[i].equals("delta_0000001_0000004")) {
                resultDelta = stat[i].getPath();
            }
        }
        Arrays.sort(names);
        String[] expected = new String[] { "delta_0000001_0000002", "delta_0000001_0000004", "delta_0000003_0000004" };
        if (!Arrays.deepEquals(expected, names)) {
            Assert.fail("Expected: " + Arrays.toString(expected) + ", found: " + Arrays.toString(names));
        }
        checkExpectedTxnsPresent(null, new Path[] { resultDelta }, columnNamesProperty, columnTypesProperty, 0, 1L, 4L, 1);
    } finally {
        connection.close();
    }
}
Also used : Path(org.apache.hadoop.fs.Path) HiveMetaStoreClient(org.apache.hadoop.hive.metastore.HiveMetaStoreClient) Table(org.apache.hadoop.hive.metastore.api.Table) FileStatus(org.apache.hadoop.fs.FileStatus) TransactionBatch(org.apache.hive.hcatalog.streaming.TransactionBatch) StreamingConnection(org.apache.hive.hcatalog.streaming.StreamingConnection) IMetaStoreClient(org.apache.hadoop.hive.metastore.IMetaStoreClient) HiveEndPoint(org.apache.hive.hcatalog.streaming.HiveEndPoint) AtomicBoolean(java.util.concurrent.atomic.AtomicBoolean) FileSystem(org.apache.hadoop.fs.FileSystem) DelimitedInputWriter(org.apache.hive.hcatalog.streaming.DelimitedInputWriter) HiveEndPoint(org.apache.hive.hcatalog.streaming.HiveEndPoint) TxnStore(org.apache.hadoop.hive.metastore.txn.TxnStore) CompactionRequest(org.apache.hadoop.hive.metastore.api.CompactionRequest) Test(org.junit.Test)

Example 38 with IMetaStoreClient

use of org.apache.hadoop.hive.metastore.IMetaStoreClient in project hive by apache.

the class TestCompactor method majorCompactAfterAbort.

@Test
public void majorCompactAfterAbort() throws Exception {
    String dbName = "default";
    String tblName = "cws";
    List<String> colNames = Arrays.asList("a", "b");
    String columnNamesProperty = "a,b";
    String columnTypesProperty = "int:string";
    executeStatementOnDriver("drop table if exists " + tblName, driver);
    executeStatementOnDriver("CREATE TABLE " + tblName + "(a INT, b STRING) " + // currently ACID requires table to be bucketed
    " CLUSTERED BY(a) INTO 1 BUCKETS" + " STORED AS ORC  TBLPROPERTIES ('transactional'='true')", driver);
    HiveEndPoint endPt = new HiveEndPoint(null, dbName, tblName, null);
    DelimitedInputWriter writer = new DelimitedInputWriter(new String[] { "a", "b" }, ",", endPt);
    StreamingConnection connection = endPt.newConnection(false, "UT_" + Thread.currentThread().getName());
    try {
        // Write a couple of batches
        for (int i = 0; i < 2; i++) {
            writeBatch(connection, writer, false);
        }
        // Start a third batch, but don't close it.
        TransactionBatch txnBatch = connection.fetchTransactionBatch(2, writer);
        txnBatch.beginNextTransaction();
        txnBatch.abort();
        txnBatch.beginNextTransaction();
        txnBatch.abort();
        // Now, compact
        TxnStore txnHandler = TxnUtils.getTxnStore(conf);
        txnHandler.compact(new CompactionRequest(dbName, tblName, CompactionType.MAJOR));
        Worker t = new Worker();
        t.setThreadId((int) t.getId());
        t.setConf(conf);
        AtomicBoolean stop = new AtomicBoolean(true);
        AtomicBoolean looped = new AtomicBoolean();
        t.init(stop, looped);
        t.run();
        // Find the location of the table
        IMetaStoreClient msClient = new HiveMetaStoreClient(conf);
        Table table = msClient.getTable(dbName, tblName);
        FileSystem fs = FileSystem.get(conf);
        FileStatus[] stat = fs.listStatus(new Path(table.getSd().getLocation()), AcidUtils.baseFileFilter);
        if (1 != stat.length) {
            Assert.fail("majorCompactAfterAbort FileStatus[] stat " + Arrays.toString(stat));
        }
        if (1 != stat.length) {
            Assert.fail("Expecting 1 file \"base_0000004\" and found " + stat.length + " files " + Arrays.toString(stat));
        }
        String name = stat[0].getPath().getName();
        if (!name.equals("base_0000004")) {
            Assert.fail("majorCompactAfterAbort name " + name + " not equals to base_0000004");
        }
        checkExpectedTxnsPresent(stat[0].getPath(), null, columnNamesProperty, columnTypesProperty, 0, 1L, 4L, 1);
    } finally {
        connection.close();
    }
}
Also used : Path(org.apache.hadoop.fs.Path) HiveMetaStoreClient(org.apache.hadoop.hive.metastore.HiveMetaStoreClient) Table(org.apache.hadoop.hive.metastore.api.Table) FileStatus(org.apache.hadoop.fs.FileStatus) TransactionBatch(org.apache.hive.hcatalog.streaming.TransactionBatch) StreamingConnection(org.apache.hive.hcatalog.streaming.StreamingConnection) IMetaStoreClient(org.apache.hadoop.hive.metastore.IMetaStoreClient) HiveEndPoint(org.apache.hive.hcatalog.streaming.HiveEndPoint) AtomicBoolean(java.util.concurrent.atomic.AtomicBoolean) FileSystem(org.apache.hadoop.fs.FileSystem) DelimitedInputWriter(org.apache.hive.hcatalog.streaming.DelimitedInputWriter) HiveEndPoint(org.apache.hive.hcatalog.streaming.HiveEndPoint) TxnStore(org.apache.hadoop.hive.metastore.txn.TxnStore) CompactionRequest(org.apache.hadoop.hive.metastore.api.CompactionRequest) Test(org.junit.Test)

Example 39 with IMetaStoreClient

use of org.apache.hadoop.hive.metastore.IMetaStoreClient in project hive by apache.

the class SecureProxySupport method buildHcatDelegationToken.

private String buildHcatDelegationToken(String user) throws IOException, InterruptedException, TException {
    final HiveConf c = new HiveConf();
    final IMetaStoreClient client = HCatUtil.getHiveMetastoreClient(c);
    LOG.info("user: " + user + " loginUser: " + UserGroupInformation.getLoginUser().getUserName());
    final UserGroupInformation ugi = UgiFactory.getUgi(user);
    String s = ugi.doAs(new PrivilegedExceptionAction<String>() {

        public String run() throws IOException, MetaException, TException {
            String u = ugi.getUserName();
            return client.getDelegationToken(c.getUser(), u);
        }
    });
    FileSystem.closeAllForUGI(ugi);
    return s;
}
Also used : TException(org.apache.thrift.TException) HiveConf(org.apache.hadoop.hive.conf.HiveConf) IOException(java.io.IOException) IMetaStoreClient(org.apache.hadoop.hive.metastore.IMetaStoreClient) UserGroupInformation(org.apache.hadoop.security.UserGroupInformation) MetaException(org.apache.hadoop.hive.metastore.api.MetaException)

Example 40 with IMetaStoreClient

use of org.apache.hadoop.hive.metastore.IMetaStoreClient in project hive by apache.

the class FileOutputCommitterContainer method cancelDelegationTokens.

private void cancelDelegationTokens(JobContext context) throws IOException {
    LOG.info("Cancelling delegation token for the job.");
    IMetaStoreClient client = null;
    try {
        HiveConf hiveConf = HCatUtil.getHiveConf(context.getConfiguration());
        client = HCatUtil.getHiveMetastoreClient(hiveConf);
        // cancel the deleg. tokens that were acquired for this job now that
        // we are done - we should cancel if the tokens were acquired by
        // HCatOutputFormat and not if they were supplied by Oozie.
        // In the latter case the HCAT_KEY_TOKEN_SIGNATURE property in
        // the conf will not be set
        String tokenStrForm = client.getTokenStrForm();
        String hCatKeyTokenSignature = context.getConfiguration().get(HCatConstants.HCAT_KEY_TOKEN_SIGNATURE);
        if (tokenStrForm != null && hCatKeyTokenSignature != null) {
            LOG.info("FileOutputCommitterContainer::cancelDelegationTokens(): " + "Cancelling token fetched for HCAT_KEY_TOKEN_SIGNATURE == (" + hCatKeyTokenSignature + ").");
            client.cancelDelegationToken(tokenStrForm);
        } else {
            LOG.info("FileOutputCommitterContainer::cancelDelegationTokens(): " + "Could not find tokenStrForm, or HCAT_KEY_TOKEN_SIGNATURE. Skipping token cancellation.");
        }
    } catch (MetaException e) {
        LOG.warn("MetaException while cancelling delegation token.", e);
    } catch (TException e) {
        LOG.warn("TException while cancelling delegation token.", e);
    } finally {
        HCatUtil.closeHiveClientQuietly(client);
    }
}
Also used : TException(org.apache.thrift.TException) HiveConf(org.apache.hadoop.hive.conf.HiveConf) IMetaStoreClient(org.apache.hadoop.hive.metastore.IMetaStoreClient) MetaException(org.apache.hadoop.hive.metastore.api.MetaException)

Aggregations

IMetaStoreClient (org.apache.hadoop.hive.metastore.IMetaStoreClient)43 TException (org.apache.thrift.TException)12 IOException (java.io.IOException)11 Path (org.apache.hadoop.fs.Path)11 HiveMetaStoreClient (org.apache.hadoop.hive.metastore.HiveMetaStoreClient)11 MetaException (org.apache.hadoop.hive.metastore.api.MetaException)11 Test (org.junit.Test)11 HiveConf (org.apache.hadoop.hive.conf.HiveConf)10 Table (org.apache.hadoop.hive.metastore.api.Table)10 FileStatus (org.apache.hadoop.fs.FileStatus)9 FileSystem (org.apache.hadoop.fs.FileSystem)9 TxnStore (org.apache.hadoop.hive.metastore.txn.TxnStore)9 AtomicBoolean (java.util.concurrent.atomic.AtomicBoolean)8 CompactionRequest (org.apache.hadoop.hive.metastore.api.CompactionRequest)8 HiveEndPoint (org.apache.hive.hcatalog.streaming.HiveEndPoint)8 ArrayList (java.util.ArrayList)7 HiveSQLException (org.apache.hive.service.cli.HiveSQLException)7 DelimitedInputWriter (org.apache.hive.hcatalog.streaming.DelimitedInputWriter)6 StreamingConnection (org.apache.hive.hcatalog.streaming.StreamingConnection)6 Table (org.apache.hadoop.hive.ql.metadata.Table)5