Examples with MapredWork org.apache.hadoop.hive.ql.plan.MapredWork used on opensource projects

Example 11 with MapredWork

use of org.apache.hadoop.hive.ql.plan.MapredWork in project hive by apache.

the class GenMapRedUtils method initPlan.

/**
 * Initialize the current plan by adding it to root tasks.
 *
 * @param op
 *          the reduce sink operator encountered
 * @param opProcCtx
 *          processing context
 */
public static void initPlan(ReduceSinkOperator op, GenMRProcContext opProcCtx) throws SemanticException {
    Operator<? extends OperatorDesc> reducer = op.getChildOperators().get(0);
    Map<Operator<? extends OperatorDesc>, GenMapRedCtx> mapCurrCtx = opProcCtx.getMapCurrCtx();
    GenMapRedCtx mapredCtx = mapCurrCtx.get(op.getParentOperators().get(0));
    Task<? extends Serializable> currTask = mapredCtx.getCurrTask();
    MapredWork plan = (MapredWork) currTask.getWork();
    HashMap<Operator<? extends OperatorDesc>, Task<? extends Serializable>> opTaskMap = opProcCtx.getOpTaskMap();
    TableScanOperator currTopOp = opProcCtx.getCurrTopOp();
    opTaskMap.put(reducer, currTask);
    plan.setReduceWork(new ReduceWork());
    plan.getReduceWork().setReducer(reducer);
    ReduceSinkDesc desc = op.getConf();
    plan.getReduceWork().setNumReduceTasks(desc.getNumReducers());
    if (needsTagging(plan.getReduceWork())) {
        plan.getReduceWork().setNeedsTagging(true);
    }
    assert currTopOp != null;
    String currAliasId = opProcCtx.getCurrAliasId();
    if (!opProcCtx.isSeenOp(currTask, currTopOp)) {
        setTaskPlan(currAliasId, currTopOp, currTask, false, opProcCtx);
    }
    currTopOp = null;
    currAliasId = null;
    opProcCtx.setCurrTask(currTask);
    opProcCtx.setCurrTopOp(currTopOp);
    opProcCtx.setCurrAliasId(currAliasId);
}

Example 12 with MapredWork

use of org.apache.hadoop.hive.ql.plan.MapredWork in project hive by apache.

the class GenMapRedUtils method splitTasks.

@SuppressWarnings("nls")
private static /**
 * Split two tasks by creating a temporary file between them.
 *
 * @param op reduce sink operator being processed
 * @param parentTask the parent task
 * @param childTask the child task
 * @param opProcCtx context
 */
void splitTasks(ReduceSinkOperator op, Task<? extends Serializable> parentTask, Task<? extends Serializable> childTask, GenMRProcContext opProcCtx) throws SemanticException {
    if (op.getNumParent() != 1) {
        throw new IllegalStateException("Expecting operator " + op + " to have one parent. " + "But found multiple parents : " + op.getParentOperators());
    }
    ParseContext parseCtx = opProcCtx.getParseCtx();
    parentTask.addDependentTask(childTask);
    // Root Task cannot depend on any other task, therefore childTask cannot be
    // a root Task
    List<Task<? extends Serializable>> rootTasks = opProcCtx.getRootTasks();
    if (rootTasks.contains(childTask)) {
        rootTasks.remove(childTask);
    }
    // Generate the temporary file name
    Context baseCtx = parseCtx.getContext();
    Path taskTmpDir = baseCtx.getMRTmpPath();
    Operator<? extends OperatorDesc> parent = op.getParentOperators().get(0);
    TableDesc tt_desc = PlanUtils.getIntermediateFileTableDesc(PlanUtils.getFieldSchemasFromRowSchema(parent.getSchema(), "temporarycol"));
    // Create the temporary file, its corresponding FileSinkOperaotr, and
    // its corresponding TableScanOperator.
    TableScanOperator tableScanOp = createTemporaryFile(parent, op, taskTmpDir, tt_desc, parseCtx);
    Map<Operator<? extends OperatorDesc>, GenMapRedCtx> mapCurrCtx = opProcCtx.getMapCurrCtx();
    mapCurrCtx.put(tableScanOp, new GenMapRedCtx(childTask, null));
    String streamDesc = taskTmpDir.toUri().toString();
    MapredWork cplan = (MapredWork) childTask.getWork();
    if (needsTagging(cplan.getReduceWork())) {
        Operator<? extends OperatorDesc> reducerOp = cplan.getReduceWork().getReducer();
        String id = null;
        if (reducerOp instanceof JoinOperator) {
            if (parseCtx.getJoinOps().contains(reducerOp)) {
                id = ((JoinOperator) reducerOp).getConf().getId();
            }
        } else if (reducerOp instanceof MapJoinOperator) {
            if (parseCtx.getMapJoinOps().contains(reducerOp)) {
                id = ((MapJoinOperator) reducerOp).getConf().getId();
            }
        } else if (reducerOp instanceof SMBMapJoinOperator) {
            if (parseCtx.getSmbMapJoinOps().contains(reducerOp)) {
                id = ((SMBMapJoinOperator) reducerOp).getConf().getId();
            }
        }
        if (id != null) {
            streamDesc = id + ":$INTNAME";
        } else {
            streamDesc = "$INTNAME";
        }
        String origStreamDesc = streamDesc;
        int pos = 0;
        while (cplan.getMapWork().getAliasToWork().get(streamDesc) != null) {
            streamDesc = origStreamDesc.concat(String.valueOf(++pos));
        }
        // TODO: Allocate work to remove the temporary files and make that
        // dependent on the redTask
        cplan.getReduceWork().setNeedsTagging(true);
    }
    // Add the path to alias mapping
    setTaskPlan(taskTmpDir, streamDesc, tableScanOp, cplan.getMapWork(), false, tt_desc);
    opProcCtx.setCurrTopOp(null);
    opProcCtx.setCurrAliasId(null);
    opProcCtx.setCurrTask(childTask);
    opProcCtx.addRootIfPossible(parentTask);
}

Example 13 with MapredWork

use of org.apache.hadoop.hive.ql.plan.MapredWork in project hive by apache.

the class GenMapRedUtils method getMapRedWorkFromConf.

/**
 * create a new plan and return. The pan won't contain the name to split
 * sample information in parse context.
 *
 * @return the new plan
 */
public static MapredWork getMapRedWorkFromConf(HiveConf conf) {
    MapredWork mrWork = new MapredWork();
    MapWork work = mrWork.getMapWork();
    boolean mapperCannotSpanPartns = conf.getBoolVar(HiveConf.ConfVars.HIVE_MAPPER_CANNOT_SPAN_MULTIPLE_PARTITIONS);
    work.setMapperCannotSpanPartns(mapperCannotSpanPartns);
    work.setPathToAliases(new LinkedHashMap<Path, ArrayList<String>>());
    work.setPathToPartitionInfo(new LinkedHashMap<Path, PartitionDesc>());
    work.setAliasToWork(new LinkedHashMap<String, Operator<? extends OperatorDesc>>());
    return mrWork;
}

Example 14 with MapredWork

use of org.apache.hadoop.hive.ql.plan.MapredWork in project hive by apache.

the class GenMapRedUtils method createMRWorkForMergingFiles.

/**
 * @param fsInput The FileSink operator.
 * @param finalName the final destination path the merge job should output.
 * @param dependencyTask
 * @param mvTasks
 * @param conf
 * @param currTask
 * @param lineageState
 * @throws SemanticException
 *
 * create a Map-only merge job using CombineHiveInputFormat for all partitions with
 * following operators:
 *          MR job J0:
 *          ...
 *          |
 *          v
 *          FileSinkOperator_1 (fsInput)
 *          |
 *          v
 *          Merge job J1:
 *          |
 *          v
 *          TableScan (using CombineHiveInputFormat) (tsMerge)
 *          |
 *          v
 *          FileSinkOperator (fsMerge)
 *
 *          Here the pathToPartitionInfo & pathToAlias will remain the same, which means the paths
 *          do
 *          not contain the dynamic partitions (their parent). So after the dynamic partitions are
 *          created (after the first job finished before the moveTask or ConditionalTask start),
 *          we need to change the pathToPartitionInfo & pathToAlias to include the dynamic
 *          partition
 *          directories.
 */
public static void createMRWorkForMergingFiles(FileSinkOperator fsInput, Path finalName, DependencyCollectionTask dependencyTask, List<Task<MoveWork>> mvTasks, HiveConf conf, Task<? extends Serializable> currTask, LineageState lineageState) throws SemanticException {
    // 
    // 1. create the operator tree
    // 
    FileSinkDesc fsInputDesc = fsInput.getConf();
    if (Utilities.FILE_OP_LOGGER.isTraceEnabled()) {
        Utilities.FILE_OP_LOGGER.trace("Creating merge work from " + System.identityHashCode(fsInput) + " with write ID " + (fsInputDesc.isMmTable() ? fsInputDesc.getTableWriteId() : null) + " into " + finalName);
    }
    boolean isBlockMerge = (conf.getBoolVar(ConfVars.HIVEMERGERCFILEBLOCKLEVEL) && fsInputDesc.getTableInfo().getInputFileFormatClass().equals(RCFileInputFormat.class)) || (conf.getBoolVar(ConfVars.HIVEMERGEORCFILESTRIPELEVEL) && fsInputDesc.getTableInfo().getInputFileFormatClass().equals(OrcInputFormat.class));
    RowSchema inputRS = fsInput.getSchema();
    Long srcMmWriteId = fsInputDesc.isMmTable() ? fsInputDesc.getTableWriteId() : null;
    FileSinkDesc fsOutputDesc = null;
    TableScanOperator tsMerge = null;
    if (!isBlockMerge) {
        // Create a TableScan operator
        tsMerge = GenMapRedUtils.createTemporaryTableScanOperator(fsInput.getCompilationOpContext(), inputRS);
        // Create a FileSink operator
        TableDesc ts = (TableDesc) fsInputDesc.getTableInfo().clone();
        Path mergeDest = srcMmWriteId == null ? finalName : finalName.getParent();
        fsOutputDesc = new FileSinkDesc(mergeDest, ts, conf.getBoolVar(ConfVars.COMPRESSRESULT));
        fsOutputDesc.setMmWriteId(srcMmWriteId);
        fsOutputDesc.setIsMerge(true);
        // Create and attach the filesink for the merge.
        OperatorFactory.getAndMakeChild(fsOutputDesc, inputRS, tsMerge);
    }
    // If the input FileSinkOperator is a dynamic partition enabled, the tsMerge input schema
    // needs to include the partition column, and the fsOutput should have
    // a DynamicPartitionCtx to indicate that it needs to dynamically partitioned.
    DynamicPartitionCtx dpCtx = fsInputDesc.getDynPartCtx();
    if (dpCtx != null && dpCtx.getNumDPCols() > 0) {
        // adding DP ColumnInfo to the RowSchema signature
        ArrayList<ColumnInfo> signature = inputRS.getSignature();
        String tblAlias = fsInputDesc.getTableInfo().getTableName();
        for (String dpCol : dpCtx.getDPColNames()) {
            ColumnInfo colInfo = new ColumnInfo(dpCol, // all partition column type should be string
            TypeInfoFactory.stringTypeInfo, tblAlias, // partition column is virtual column
            true);
            signature.add(colInfo);
        }
        inputRS.setSignature(signature);
        if (!isBlockMerge) {
            // create another DynamicPartitionCtx, which has a different input-to-DP column mapping
            DynamicPartitionCtx dpCtx2 = new DynamicPartitionCtx(dpCtx);
            fsOutputDesc.setDynPartCtx(dpCtx2);
        }
        // update the FileSinkOperator to include partition columns
        usePartitionColumns(fsInputDesc.getTableInfo().getProperties(), dpCtx.getDPColNames());
    } else {
        // non-partitioned table
        fsInputDesc.getTableInfo().getProperties().remove(org.apache.hadoop.hive.metastore.api.hive_metastoreConstants.META_TABLE_PARTITION_COLUMNS);
    }
    // 
    // 2. Constructing a conditional task consisting of a move task and a map reduce task
    // 
    Path inputDirName = fsInputDesc.getMergeInputDirName();
    MapWork cplan;
    Serializable work;
    if (isBlockMerge) {
        cplan = GenMapRedUtils.createMergeTask(fsInputDesc, finalName, dpCtx != null && dpCtx.getNumDPCols() > 0, fsInput.getCompilationOpContext());
        if (conf.getVar(ConfVars.HIVE_EXECUTION_ENGINE).equals("tez")) {
            work = new TezWork(conf.getVar(HiveConf.ConfVars.HIVEQUERYID), conf);
            cplan.setName("File Merge");
            ((TezWork) work).add(cplan);
        } else if (conf.getVar(ConfVars.HIVE_EXECUTION_ENGINE).equals("spark")) {
            work = new SparkWork(conf.getVar(HiveConf.ConfVars.HIVEQUERYID));
            cplan.setName("Spark Merge File Work");
            ((SparkWork) work).add(cplan);
        } else {
            work = cplan;
        }
    } else {
        cplan = createMRWorkForMergingFiles(conf, tsMerge, fsInputDesc);
        if (conf.getVar(ConfVars.HIVE_EXECUTION_ENGINE).equals("tez")) {
            work = new TezWork(conf.getVar(HiveConf.ConfVars.HIVEQUERYID), conf);
            cplan.setName("File Merge");
            ((TezWork) work).add(cplan);
        } else if (conf.getVar(ConfVars.HIVE_EXECUTION_ENGINE).equals("spark")) {
            work = new SparkWork(conf.getVar(HiveConf.ConfVars.HIVEQUERYID));
            cplan.setName("Spark Merge File Work");
            ((SparkWork) work).add(cplan);
        } else {
            work = new MapredWork();
            ((MapredWork) work).setMapWork(cplan);
        }
    }
    // use CombineHiveInputFormat for map-only merging
    cplan.setInputformat("org.apache.hadoop.hive.ql.io.CombineHiveInputFormat");
    // NOTE: we should gather stats in MR1 rather than MR2 at merge job since we don't
    // know if merge MR2 will be triggered at execution time
    MoveWork dummyMv = null;
    if (srcMmWriteId == null) {
        // Only create the movework for non-MM table. No action needed for a MM table.
        dummyMv = new MoveWork(null, null, null, new LoadFileDesc(inputDirName, finalName, true, null, null, false), false);
    }
    // Use the original fsOp path here in case of MM - while the new FSOP merges files inside the
    // MM directory, the original MoveTask still commits based on the parent. Note that this path
    // can only be triggered for a merge that's part of insert for now; MM tables do not support
    // concatenate. Keeping the old logic for non-MM tables with temp directories and stuff.
    Path fsopPath = srcMmWriteId != null ? fsInputDesc.getFinalDirName() : finalName;
    Task<MoveWork> mvTask = GenMapRedUtils.findMoveTaskForFsopOutput(mvTasks, fsopPath, fsInputDesc.isMmTable());
    ConditionalTask cndTsk = GenMapRedUtils.createCondTask(conf, currTask, dummyMv, work, fsInputDesc.getMergeInputDirName(), finalName, mvTask, dependencyTask, lineageState);
    // keep the dynamic partition context in conditional task resolver context
    ConditionalResolverMergeFilesCtx mrCtx = (ConditionalResolverMergeFilesCtx) cndTsk.getResolverCtx();
    mrCtx.setDPCtx(fsInputDesc.getDynPartCtx());
    mrCtx.setLbCtx(fsInputDesc.getLbCtx());
}

Example 15 with MapredWork

use of org.apache.hadoop.hive.ql.plan.MapredWork in project hive by apache.

the class GenMapRedUtils method splitPlan.

/**
 * Met cRS in pOP(parentTask with RS)-cRS-cOP(noTask) case
 * Create new child task for cRS-cOP and link two tasks by temporary file : pOP-FS / TS-cRS-cOP
 *
 * @param cRS
 *          the reduce sink operator encountered
 * @param opProcCtx
 *          processing context
 */
static void splitPlan(ReduceSinkOperator cRS, GenMRProcContext opProcCtx) throws SemanticException {
    // Generate a new task
    ParseContext parseCtx = opProcCtx.getParseCtx();
    Task<? extends Serializable> parentTask = opProcCtx.getCurrTask();
    MapredWork childPlan = getMapRedWork(parseCtx);
    Task<? extends Serializable> childTask = TaskFactory.get(childPlan);
    Operator<? extends OperatorDesc> reducer = cRS.getChildOperators().get(0);
    // Add the reducer
    ReduceWork rWork = new ReduceWork();
    childPlan.setReduceWork(rWork);
    rWork.setReducer(reducer);
    ReduceSinkDesc desc = cRS.getConf();
    childPlan.getReduceWork().setNumReduceTasks(new Integer(desc.getNumReducers()));
    opProcCtx.getOpTaskMap().put(reducer, childTask);
    splitTasks(cRS, parentTask, childTask, opProcCtx);
}