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

Example 41 with ShowCompactRequest

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

the class TestTxnCommands2 method testCompactWithDelete.

@Test
public void testCompactWithDelete() throws Exception {
    int[][] tableData = { { 1, 2 }, { 3, 4 } };
    runStatementOnDriver("insert into " + Table.ACIDTBL + "(a,b) " + makeValuesClause(tableData));
    runStatementOnDriver("alter table " + Table.ACIDTBL + " compact 'MAJOR'");
    Worker t = new Worker();
    t.setThreadId((int) t.getId());
    t.setConf(hiveConf);
    AtomicBoolean stop = new AtomicBoolean();
    AtomicBoolean looped = new AtomicBoolean();
    stop.set(true);
    t.init(stop, looped);
    t.run();
    runStatementOnDriver("delete from " + Table.ACIDTBL + " where b = 4");
    runStatementOnDriver("update " + Table.ACIDTBL + " set b = -2 where b = 2");
    runStatementOnDriver("alter table " + Table.ACIDTBL + " compact 'MINOR'");
    t.run();
    TxnStore txnHandler = TxnUtils.getTxnStore(hiveConf);
    ShowCompactResponse resp = txnHandler.showCompact(new ShowCompactRequest());
    Assert.assertEquals("Unexpected number of compactions in history", 2, resp.getCompactsSize());
    Assert.assertEquals("Unexpected 0 compaction state", TxnStore.CLEANING_RESPONSE, resp.getCompacts().get(0).getState());
    Assert.assertEquals("Unexpected 1 compaction state", TxnStore.CLEANING_RESPONSE, resp.getCompacts().get(1).getState());
}

Example 42 with ShowCompactRequest

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

the class TestTxnCommands2 method testEmptyInTblproperties.

/**
 * https://issues.apache.org/jira/browse/HIVE-17391
 */
@Test
public void testEmptyInTblproperties() throws Exception {
    runStatementOnDriver("create table t1 " + "(a int, b int) stored as orc TBLPROPERTIES ('serialization.null.format'='', 'transactional'='true')");
    runStatementOnDriver("insert into t1 " + "(a,b) values(1,7),(3,7)");
    runStatementOnDriver("update t1" + " set b = -2 where b = 2");
    runStatementOnDriver("alter table t1 " + " compact 'MAJOR'");
    runWorker(hiveConf);
    TxnStore txnHandler = TxnUtils.getTxnStore(hiveConf);
    ShowCompactResponse resp = txnHandler.showCompact(new ShowCompactRequest());
    Assert.assertEquals("Unexpected number of compactions in history", 1, resp.getCompactsSize());
    Assert.assertEquals("Unexpected 0 compaction state", TxnStore.CLEANING_RESPONSE, resp.getCompacts().get(0).getState());
    Assert.assertTrue(resp.getCompacts().get(0).getHadoopJobId().startsWith("job_local"));
}

Example 43 with ShowCompactRequest

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

the class TestTxnNoBuckets method testEmptyCompactionResult.

/**
 * see HIVE-18429
 */
@Test
public void testEmptyCompactionResult() throws Exception {
    hiveConf.set(MetastoreConf.ConfVars.CREATE_TABLES_AS_ACID.getVarname(), "true");
    runStatementOnDriver("drop table if exists T");
    runStatementOnDriver("create table T (a int, b int) stored as orc");
    int[][] data = { { 1, 2 }, { 3, 4 } };
    runStatementOnDriver("insert into T" + makeValuesClause(data));
    runStatementOnDriver("insert into T" + makeValuesClause(data));
    // delete the bucket files so now we have empty delta dirs
    List<String> rs = runStatementOnDriver("select distinct INPUT__FILE__NAME from T");
    FileSystem fs = FileSystem.get(hiveConf);
    for (String path : rs) {
        fs.delete(new Path(path), true);
    }
    runStatementOnDriver("alter table T compact 'major'");
    TestTxnCommands2.runWorker(hiveConf);
    // check status of compaction job
    TxnStore txnHandler = TxnUtils.getTxnStore(hiveConf);
    ShowCompactResponse resp = txnHandler.showCompact(new ShowCompactRequest());
    Assert.assertEquals("Unexpected number of compactions in history", 1, resp.getCompactsSize());
    Assert.assertEquals("Unexpected 0 compaction state", TxnStore.CLEANING_RESPONSE, resp.getCompacts().get(0).getState());
    Assert.assertTrue(resp.getCompacts().get(0).getHadoopJobId().startsWith("job_local"));
    // now run another compaction make sure empty dirs don't cause issues
    runStatementOnDriver("insert into T" + makeValuesClause(data));
    runStatementOnDriver("alter table T compact 'major'");
    TestTxnCommands2.runWorker(hiveConf);
    // check status of compaction job
    resp = txnHandler.showCompact(new ShowCompactRequest());
    Assert.assertEquals("Unexpected number of compactions in history", 2, resp.getCompactsSize());
    for (int i = 0; i < 2; i++) {
        Assert.assertEquals("Unexpected 0 compaction state", TxnStore.CLEANING_RESPONSE, resp.getCompacts().get(i).getState());
        Assert.assertTrue(resp.getCompacts().get(i).getHadoopJobId().startsWith("job_local"));
    }
    rs = runStatementOnDriver("select a, b from T order by a, b");
    Assert.assertEquals(stringifyValues(data), rs);
}

Example 44 with ShowCompactRequest

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

the class TestTxnNoBuckets method testNonAcidToAcidVectorzied.

/**
 * Tests to check that we are able to use vectorized acid reader,
 * VectorizedOrcAcidRowBatchReader, when reading "original" files,
 * i.e. those that were written before the table was converted to acid.
 * See also acid_vectorization_original*.q
 */
@Test
public void testNonAcidToAcidVectorzied() throws Exception {
    hiveConf.setBoolVar(HiveConf.ConfVars.HIVE_VECTORIZATION_ENABLED, true);
    hiveConf.setVar(HiveConf.ConfVars.HIVEFETCHTASKCONVERSION, "none");
    // this enables vectorization of ROW__ID
    // HIVE-12631
    hiveConf.setBoolVar(HiveConf.ConfVars.HIVE_VECTORIZATION_ROW_IDENTIFIER_ENABLED, true);
    runStatementOnDriver("drop table if exists T");
    runStatementOnDriver("create table T(a int, b int) stored as orc tblproperties('transactional'='false')");
    int[][] values = { { 1, 2 }, { 2, 4 }, { 5, 6 }, { 6, 8 }, { 9, 10 } };
    runStatementOnDriver("insert into T(a, b) " + makeValuesClause(values));
    // , 'transactional_properties'='default'
    runStatementOnDriver("alter table T SET TBLPROPERTIES ('transactional'='true')");
    // Execution mode: vectorized
    // this uses VectorizedOrcAcidRowBatchReader
    String query = "select a from T where b > 6 order by a";
    List<String> rs = runStatementOnDriver(query);
    String[][] expected = { { "6", "" }, { "9", "" } };
    checkExpected(rs, expected, "After conversion");
    Assert.assertEquals(Integer.toString(6), rs.get(0));
    Assert.assertEquals(Integer.toString(9), rs.get(1));
    assertVectorized(shouldVectorize(), query);
    // why isn't PPD working.... - it is working but storage layer doesn't do row level filtering; only row group level
    // this uses VectorizedOrcAcidRowBatchReader
    query = "select ROW__ID, a from T where b > 6 order by a";
    rs = runStatementOnDriver(query);
    String[][] expected1 = { { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":3}", "6" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":4}", "9" } };
    checkExpected(rs, expected1, "After conversion with VC1");
    assertVectorized(shouldVectorize(), query);
    // this uses VectorizedOrcAcidRowBatchReader
    query = "select ROW__ID, a from T where b > 0 order by a";
    rs = runStatementOnDriver(query);
    String[][] expected2 = { { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":0}", "1" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":1}", "2" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":2}", "5" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":3}", "6" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":4}", "9" } };
    checkExpected(rs, expected2, "After conversion with VC2");
    assertVectorized(shouldVectorize(), query);
    // doesn't vectorize (uses neither of the Vectorzied Acid readers)
    query = "select ROW__ID, a, INPUT__FILE__NAME from T where b > 6 order by a";
    rs = runStatementOnDriver(query);
    Assert.assertEquals("", 2, rs.size());
    String[][] expected3 = { { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":3}\t6", "warehouse/t/000000_0" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":4}\t9", "warehouse/t/000000_0" } };
    checkExpected(rs, expected3, "After non-vectorized read");
    Assert.assertEquals(0, BucketCodec.determineVersion(536870912).decodeWriterId(536870912));
    // vectorized because there is INPUT__FILE__NAME
    assertVectorized(false, query);
    runStatementOnDriver("update T set b = 17 where a = 1");
    // this should use VectorizedOrcAcidRowReader
    query = "select ROW__ID, b from T where b > 0 order by a";
    rs = runStatementOnDriver(query);
    String[][] expected4 = { { "{\"writeid\":1,\"bucketid\":536870912,\"rowid\":0}", "17" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":1}", "4" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":2}", "6" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":3}", "8" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":4}", "10" } };
    checkExpected(rs, expected4, "After conversion with VC4");
    assertVectorized(shouldVectorize(), query);
    runStatementOnDriver("alter table T compact 'major'");
    TestTxnCommands2.runWorker(hiveConf);
    TxnStore txnHandler = TxnUtils.getTxnStore(hiveConf);
    ShowCompactResponse resp = txnHandler.showCompact(new ShowCompactRequest());
    Assert.assertEquals("Unexpected number of compactions in history", 1, resp.getCompactsSize());
    Assert.assertEquals("Unexpected 0 compaction state", TxnStore.CLEANING_RESPONSE, resp.getCompacts().get(0).getState());
    Assert.assertTrue(resp.getCompacts().get(0).getHadoopJobId().startsWith("job_local"));
    // this should not vectorize at all
    query = "select ROW__ID, a, b, INPUT__FILE__NAME from T where b > 0 order by a, b";
    rs = runStatementOnDriver(query);
    String[][] expected5 = { // the row__ids are the same after compaction
    { "{\"writeid\":1,\"bucketid\":536870912,\"rowid\":0}\t1\t17", "warehouse/t/base_0000001/bucket_00000" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":1}\t2\t4", "warehouse/t/base_0000001/bucket_00000" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":2}\t5\t6", "warehouse/t/base_0000001/bucket_00000" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":3}\t6\t8", "warehouse/t/base_0000001/bucket_00000" }, { "{\"writeid\":0,\"bucketid\":536870912,\"rowid\":4}\t9\t10", "warehouse/t/base_0000001/bucket_00000" } };
    checkExpected(rs, expected5, "After major compaction");
    // vectorized because there is INPUT__FILE__NAME
    assertVectorized(false, query);
}

Example 45 with ShowCompactRequest

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

the class Initiator method run.

@Override
public void run() {
    // so wrap it in a big catch Throwable statement.
    try {
        recoverFailedCompactions(false);
        int abortedThreshold = HiveConf.getIntVar(conf, HiveConf.ConfVars.HIVE_COMPACTOR_ABORTEDTXN_THRESHOLD);
        // HiveMetaStore.
        do {
            long startedAt = -1;
            TxnStore.MutexAPI.LockHandle handle = null;
            // don't doom the entire thread.
            try {
                handle = txnHandler.getMutexAPI().acquireLock(TxnStore.MUTEX_KEY.Initiator.name());
                startedAt = System.currentTimeMillis();
                // todo: add method to only get current i.e. skip history - more efficient
                ShowCompactResponse currentCompactions = txnHandler.showCompact(new ShowCompactRequest());
                Set<CompactionInfo> potentials = txnHandler.findPotentialCompactions(abortedThreshold);
                LOG.debug("Found " + potentials.size() + " potential compactions, " + "checking to see if we should compact any of them");
                for (CompactionInfo ci : potentials) {
                    LOG.info("Checking to see if we should compact " + ci.getFullPartitionName());
                    try {
                        Table t = resolveTable(ci);
                        if (t == null) {
                            // Most likely this means it's a temp table
                            LOG.info("Can't find table " + ci.getFullTableName() + ", assuming it's a temp " + "table or has been dropped and moving on.");
                            continue;
                        }
                        // check if no compaction set for this table
                        if (noAutoCompactSet(t)) {
                            LOG.info("Table " + tableName(t) + " marked " + hive_metastoreConstants.TABLE_NO_AUTO_COMPACT + "=true so we will not compact it.");
                            continue;
                        }
                        // then it's a dynamic partitioning case and we shouldn't check the table itself.
                        if (t.getPartitionKeys() != null && t.getPartitionKeys().size() > 0 && ci.partName == null) {
                            LOG.debug("Skipping entry for " + ci.getFullTableName() + " as it is from dynamic" + " partitioning");
                            continue;
                        }
                        // the time currentCompactions is generated and now
                        if (lookForCurrentCompactions(currentCompactions, ci)) {
                            LOG.debug("Found currently initiated or working compaction for " + ci.getFullPartitionName() + " so we will not initiate another compaction");
                            continue;
                        }
                        if (txnHandler.checkFailedCompactions(ci)) {
                            LOG.warn("Will not initiate compaction for " + ci.getFullPartitionName() + " since last " + HiveConf.ConfVars.COMPACTOR_INITIATOR_FAILED_THRESHOLD + " attempts to compact it failed.");
                            txnHandler.markFailed(ci);
                            continue;
                        }
                        // Figure out who we should run the file operations as
                        Partition p = resolvePartition(ci);
                        if (p == null && ci.partName != null) {
                            LOG.info("Can't find partition " + ci.getFullPartitionName() + ", assuming it has been dropped and moving on.");
                            continue;
                        }
                        // Compaction doesn't work under a transaction and hence pass null for validTxnList
                        // The response will have one entry per table and hence we get only one ValidWriteIdList
                        String fullTableName = TxnUtils.getFullTableName(t.getDbName(), t.getTableName());
                        GetValidWriteIdsRequest rqst = new GetValidWriteIdsRequest(Collections.singletonList(fullTableName), null);
                        ValidWriteIdList tblValidWriteIds = TxnUtils.createValidCompactWriteIdList(txnHandler.getValidWriteIds(rqst).getTblValidWriteIds().get(0));
                        StorageDescriptor sd = resolveStorageDescriptor(t, p);
                        String runAs = findUserToRunAs(sd.getLocation(), t);
                        /*Future thought: checkForCompaction will check a lot of file metadata and may be expensive.
              * Long term we should consider having a thread pool here and running checkForCompactionS
              * in parallel*/
                        CompactionType compactionNeeded = checkForCompaction(ci, tblValidWriteIds, sd, t.getParameters(), runAs);
                        if (compactionNeeded != null)
                            requestCompaction(ci, runAs, compactionNeeded);
                    } catch (Throwable t) {
                        LOG.error("Caught exception while trying to determine if we should compact " + ci + ".  Marking failed to avoid repeated failures, " + "" + StringUtils.stringifyException(t));
                        txnHandler.markFailed(ci);
                    }
                }
                // Check for timed out remote workers.
                recoverFailedCompactions(true);
                // Clean anything from the txns table that has no components left in txn_components.
                txnHandler.cleanEmptyAbortedTxns();
            } catch (Throwable t) {
                LOG.error("Initiator loop caught unexpected exception this time through the loop: " + StringUtils.stringifyException(t));
            } finally {
                if (handle != null) {
                    handle.releaseLocks();
                }
            }
            long elapsedTime = System.currentTimeMillis() - startedAt;
            if (elapsedTime >= checkInterval || stop.get())
                continue;
            else
                Thread.sleep(checkInterval - elapsedTime);
        } while (!stop.get());
    } catch (Throwable t) {
        LOG.error("Caught an exception in the main loop of compactor initiator, exiting " + StringUtils.stringifyException(t));
    }
}