Search in sources :

Example 21 with JoinDesc

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

the class TopNKeyPushdownProcessor method pushdownThroughLeftOuterJoin.

/**
 * Push through LOJ. If TopNKey expression refers fully to expressions from left input, push
 * with rewriting of expressions and remove from top of LOJ. If TopNKey expression has a prefix
 * that refers to expressions from left input, push with rewriting of those expressions and keep
 * on top of LOJ.
 *
 * @param topNKey TopNKey operator to push
 * @throws SemanticException when removeChildAndAdoptItsChildren was not successful
 */
private void pushdownThroughLeftOuterJoin(TopNKeyOperator topNKey) throws SemanticException {
    final TopNKeyDesc topNKeyDesc = topNKey.getConf();
    final CommonJoinOperator<? extends JoinDesc> join = (CommonJoinOperator<? extends JoinDesc>) topNKey.getParentOperators().get(0);
    final List<Operator<? extends OperatorDesc>> joinInputs = join.getParentOperators();
    final ReduceSinkOperator reduceSinkOperator = (ReduceSinkOperator) joinInputs.get(0);
    final ReduceSinkDesc reduceSinkDesc = reduceSinkOperator.getConf();
    CommonKeyPrefix commonKeyPrefix = CommonKeyPrefix.map(mapUntilColumnEquals(topNKeyDesc.getKeyColumns(), join.getColumnExprMap()), topNKeyDesc.getColumnSortOrder(), topNKeyDesc.getNullOrder(), reduceSinkDesc.getKeyCols(), reduceSinkDesc.getColumnExprMap(), reduceSinkDesc.getOrder(), reduceSinkDesc.getNullOrder());
    if (commonKeyPrefix.isEmpty() || commonKeyPrefix.size() == topNKeyDesc.getPartitionKeyColumns().size()) {
        return;
    }
    LOG.debug("Pushing a copy of {} through {} and {}", topNKey.getName(), join.getName(), reduceSinkOperator.getName());
    final TopNKeyDesc newTopNKeyDesc = topNKeyDesc.combine(commonKeyPrefix);
    pushdown((TopNKeyOperator) copyDown(reduceSinkOperator, newTopNKeyDesc));
    if (topNKeyDesc.getKeyColumns().size() == commonKeyPrefix.size()) {
        LOG.debug("Removing {} above {}", topNKey.getName(), join.getName());
        join.removeChildAndAdoptItsChildren(topNKey);
    }
}
Also used : ReduceSinkOperator(org.apache.hadoop.hive.ql.exec.ReduceSinkOperator) CommonJoinOperator(org.apache.hadoop.hive.ql.exec.CommonJoinOperator) GroupByOperator(org.apache.hadoop.hive.ql.exec.GroupByOperator) TopNKeyOperator(org.apache.hadoop.hive.ql.exec.TopNKeyOperator) Operator(org.apache.hadoop.hive.ql.exec.Operator) SelectOperator(org.apache.hadoop.hive.ql.exec.SelectOperator) TopNKeyDesc(org.apache.hadoop.hive.ql.plan.TopNKeyDesc) ReduceSinkOperator(org.apache.hadoop.hive.ql.exec.ReduceSinkOperator) CommonJoinOperator(org.apache.hadoop.hive.ql.exec.CommonJoinOperator) ReduceSinkDesc(org.apache.hadoop.hive.ql.plan.ReduceSinkDesc) JoinDesc(org.apache.hadoop.hive.ql.plan.JoinDesc) OperatorDesc(org.apache.hadoop.hive.ql.plan.OperatorDesc)

Example 22 with JoinDesc

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

the class TopNKeyPushdownProcessor method pushdownInnerJoin.

/**
 * Tries to push the TopNKeyFilter through an inner join:
 *  requirements:
 *    - being PK-FK join
 *    - PK side is not filtered
 *    - First n TopNKey key columns (Order By) are originated from the FK side.
 * @throws SemanticException
 */
private void pushdownInnerJoin(TopNKeyOperator topNKey, int fkJoinInputIndex, boolean nonFkSideIsFiltered) throws SemanticException {
    TopNKeyDesc topNKeyDesc = topNKey.getConf();
    CommonJoinOperator<? extends JoinDesc> join = (CommonJoinOperator<? extends JoinDesc>) topNKey.getParentOperators().get(0);
    List<Operator<? extends OperatorDesc>> joinInputs = join.getParentOperators();
    ReduceSinkOperator fkJoinInput = (ReduceSinkOperator) joinInputs.get(fkJoinInputIndex);
    if (nonFkSideIsFiltered) {
        LOG.debug("Not pushing {} through {} as non FK side of the join is filtered", topNKey.getName(), join.getName());
        return;
    }
    CommonKeyPrefix commonKeyPrefix = CommonKeyPrefix.map(mapUntilColumnEquals(topNKeyDesc.getKeyColumns(), join.getColumnExprMap()), topNKeyDesc.getColumnSortOrder(), topNKeyDesc.getNullOrder(), fkJoinInput.getConf().getKeyCols(), fkJoinInput.getConf().getColumnExprMap(), fkJoinInput.getConf().getOrder(), fkJoinInput.getConf().getNullOrder());
    if (commonKeyPrefix.isEmpty() || commonKeyPrefix.size() == topNKeyDesc.getPartitionKeyColumns().size()) {
        return;
    }
    LOG.debug("Pushing a copy of {} through {} and {}", topNKey.getName(), join.getName(), fkJoinInput.getName());
    final TopNKeyDesc newTopNKeyDesc = topNKeyDesc.combine(commonKeyPrefix);
    pushdown((TopNKeyOperator) copyDown(fkJoinInput, newTopNKeyDesc));
    if (topNKeyDesc.getKeyColumns().size() == commonKeyPrefix.size()) {
        LOG.debug("Removing {} above {}", topNKey.getName(), join.getName());
        join.removeChildAndAdoptItsChildren(topNKey);
    }
}
Also used : ReduceSinkOperator(org.apache.hadoop.hive.ql.exec.ReduceSinkOperator) CommonJoinOperator(org.apache.hadoop.hive.ql.exec.CommonJoinOperator) GroupByOperator(org.apache.hadoop.hive.ql.exec.GroupByOperator) TopNKeyOperator(org.apache.hadoop.hive.ql.exec.TopNKeyOperator) Operator(org.apache.hadoop.hive.ql.exec.Operator) SelectOperator(org.apache.hadoop.hive.ql.exec.SelectOperator) TopNKeyDesc(org.apache.hadoop.hive.ql.plan.TopNKeyDesc) ReduceSinkOperator(org.apache.hadoop.hive.ql.exec.ReduceSinkOperator) CommonJoinOperator(org.apache.hadoop.hive.ql.exec.CommonJoinOperator) JoinDesc(org.apache.hadoop.hive.ql.plan.JoinDesc) OperatorDesc(org.apache.hadoop.hive.ql.plan.OperatorDesc)

Example 23 with JoinDesc

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

the class CommonJoinTaskDispatcher method processCurrentTask.

@Override
public Task<?> processCurrentTask(MapRedTask currTask, ConditionalTask conditionalTask, Context context) throws SemanticException {
    // whether it contains common join op; if contains, return this common join op
    JoinOperator joinOp = getJoinOp(currTask);
    if (joinOp == null || joinOp.getConf().isFixedAsSorted()) {
        return null;
    }
    currTask.setTaskTag(Task.COMMON_JOIN);
    MapWork currWork = currTask.getWork().getMapWork();
    // create conditional work list and task list
    List<Serializable> listWorks = new ArrayList<Serializable>();
    List<Task<?>> listTasks = new ArrayList<Task<?>>();
    // create task to aliases mapping and alias to input file mapping for resolver
    // Must be deterministic order map for consistent q-test output across Java versions
    HashMap<Task<?>, Set<String>> taskToAliases = new LinkedHashMap<Task<?>, Set<String>>();
    Map<Path, List<String>> pathToAliases = currWork.getPathToAliases();
    Map<String, Operator<? extends OperatorDesc>> aliasToWork = currWork.getAliasToWork();
    // start to generate multiple map join tasks
    JoinDesc joinDesc = joinOp.getConf();
    if (aliasToSize == null) {
        aliasToSize = new HashMap<String, Long>();
    }
    try {
        long aliasTotalKnownInputSize = getTotalKnownInputSize(context, currWork, pathToAliases, aliasToSize);
        Set<Integer> bigTableCandidates = MapJoinProcessor.getBigTableCandidates(joinDesc.getConds());
        // no table could be the big table; there is no need to convert
        if (bigTableCandidates.isEmpty()) {
            return null;
        }
        // if any of bigTableCandidates is from multi-sourced, bigTableCandidates should
        // only contain multi-sourced because multi-sourced cannot be hashed or direct readable
        bigTableCandidates = multiInsertBigTableCheck(joinOp, bigTableCandidates);
        Configuration conf = context.getConf();
        // If sizes of at least n-1 tables in a n-way join is known, and their sum is smaller than
        // the threshold size, convert the join into map-join and don't create a conditional task
        boolean convertJoinMapJoin = HiveConf.getBoolVar(conf, HiveConf.ConfVars.HIVECONVERTJOINNOCONDITIONALTASK);
        int bigTablePosition = -1;
        if (convertJoinMapJoin) {
            // This is the threshold that the user has specified to fit in mapjoin
            long mapJoinSize = HiveConf.getLongVar(conf, HiveConf.ConfVars.HIVECONVERTJOINNOCONDITIONALTASKTHRESHOLD);
            Long bigTableSize = null;
            Set<String> aliases = aliasToWork.keySet();
            for (int tablePosition : bigTableCandidates) {
                Operator<?> parent = joinOp.getParentOperators().get(tablePosition);
                Set<String> participants = GenMapRedUtils.findAliases(currWork, parent);
                long sumOfOthers = Utilities.sumOfExcept(aliasToSize, aliases, participants);
                if (sumOfOthers < 0 || sumOfOthers > mapJoinSize) {
                    // some small alias is not known or too big
                    continue;
                }
                if (bigTableSize == null && bigTablePosition >= 0 && tablePosition < bigTablePosition) {
                    // prefer right most alias
                    continue;
                }
                long aliasSize = Utilities.sumOf(aliasToSize, participants);
                if (bigTableSize == null || bigTableSize < 0 || (aliasSize >= 0 && aliasSize >= bigTableSize)) {
                    bigTablePosition = tablePosition;
                    bigTableSize = aliasSize;
                }
            }
        }
        currWork.setLeftInputJoin(joinOp.getConf().isLeftInputJoin());
        currWork.setBaseSrc(joinOp.getConf().getBaseSrc());
        currWork.setMapAliases(joinOp.getConf().getMapAliases());
        if (bigTablePosition >= 0) {
            // create map join task and set big table as bigTablePosition
            MapRedTask newTask = convertTaskToMapJoinTask(currTask.getWork(), bigTablePosition);
            newTask.setTaskTag(Task.MAPJOIN_ONLY_NOBACKUP);
            newTask.setFetchSource(currTask.isFetchSource());
            replaceTask(currTask, newTask);
            // joined with multiple small tables on different keys
            if ((newTask.getChildTasks() != null) && (newTask.getChildTasks().size() == 1)) {
                mergeMapJoinTaskIntoItsChildMapRedTask(newTask, conf);
            }
            return newTask;
        }
        long ThresholdOfSmallTblSizeSum = HiveConf.getLongVar(conf, HiveConf.ConfVars.HIVESMALLTABLESFILESIZE);
        for (int pos = 0; pos < joinOp.getNumParent(); pos++) {
            // this table cannot be big table
            if (!bigTableCandidates.contains(pos)) {
                continue;
            }
            Operator<?> startOp = joinOp.getParentOperators().get(pos);
            Set<String> aliases = GenMapRedUtils.findAliases(currWork, startOp);
            long aliasKnownSize = Utilities.sumOf(aliasToSize, aliases);
            if (cannotConvert(aliasKnownSize, aliasTotalKnownInputSize, ThresholdOfSmallTblSizeSum)) {
                continue;
            }
            MapredWork newWork = SerializationUtilities.clonePlan(currTask.getWork());
            // create map join task and set big table as i
            MapRedTask newTask = convertTaskToMapJoinTask(newWork, pos);
            // add into conditional task
            listWorks.add(newTask.getWork());
            listTasks.add(newTask);
            newTask.setTaskTag(Task.CONVERTED_MAPJOIN);
            newTask.setFetchSource(currTask.isFetchSource());
            // set up backup task
            newTask.setBackupTask(currTask);
            newTask.setBackupChildrenTasks(currTask.getChildTasks());
            // put the mapping task to aliases
            taskToAliases.put(newTask, aliases);
        }
    } catch (Exception e) {
        throw new SemanticException("Generate Map Join Task Error: " + e.getMessage(), e);
    }
    if (listTasks.isEmpty()) {
        return currTask;
    }
    // insert current common join task to conditional task
    listWorks.add(currTask.getWork());
    listTasks.add(currTask);
    // clear JoinTree and OP Parse Context
    currWork.setLeftInputJoin(false);
    currWork.setBaseSrc(null);
    currWork.setMapAliases(null);
    // create conditional task and insert conditional task into task tree
    ConditionalWork cndWork = new ConditionalWork(listWorks);
    ConditionalTask cndTsk = (ConditionalTask) TaskFactory.get(cndWork);
    cndTsk.setListTasks(listTasks);
    // set resolver and resolver context
    cndTsk.setResolver(new ConditionalResolverCommonJoin());
    ConditionalResolverCommonJoinCtx resolverCtx = new ConditionalResolverCommonJoinCtx();
    resolverCtx.setPathToAliases(pathToAliases);
    resolverCtx.setAliasToKnownSize(aliasToSize);
    resolverCtx.setTaskToAliases(taskToAliases);
    resolverCtx.setCommonJoinTask(currTask);
    resolverCtx.setLocalTmpDir(context.getLocalScratchDir(false));
    resolverCtx.setHdfsTmpDir(context.getMRScratchDir());
    cndTsk.setResolverCtx(resolverCtx);
    // replace the current task with the new generated conditional task
    replaceTaskWithConditionalTask(currTask, cndTsk);
    return cndTsk;
}
Also used : JoinOperator(org.apache.hadoop.hive.ql.exec.JoinOperator) LateralViewForwardOperator(org.apache.hadoop.hive.ql.exec.LateralViewForwardOperator) FileSinkOperator(org.apache.hadoop.hive.ql.exec.FileSinkOperator) JoinOperator(org.apache.hadoop.hive.ql.exec.JoinOperator) TableScanOperator(org.apache.hadoop.hive.ql.exec.TableScanOperator) Operator(org.apache.hadoop.hive.ql.exec.Operator) Serializable(java.io.Serializable) ConditionalTask(org.apache.hadoop.hive.ql.exec.ConditionalTask) Task(org.apache.hadoop.hive.ql.exec.Task) MapRedTask(org.apache.hadoop.hive.ql.exec.mr.MapRedTask) HashSet(java.util.HashSet) Set(java.util.Set) Configuration(org.apache.hadoop.conf.Configuration) ArrayList(java.util.ArrayList) ConditionalWork(org.apache.hadoop.hive.ql.plan.ConditionalWork) LinkedHashMap(java.util.LinkedHashMap) MapRedTask(org.apache.hadoop.hive.ql.exec.mr.MapRedTask) ConditionalResolverCommonJoinCtx(org.apache.hadoop.hive.ql.plan.ConditionalResolverCommonJoin.ConditionalResolverCommonJoinCtx) MapredWork(org.apache.hadoop.hive.ql.plan.MapredWork) ConditionalTask(org.apache.hadoop.hive.ql.exec.ConditionalTask) ArrayList(java.util.ArrayList) List(java.util.List) ConditionalResolverCommonJoin(org.apache.hadoop.hive.ql.plan.ConditionalResolverCommonJoin) SemanticException(org.apache.hadoop.hive.ql.parse.SemanticException) Path(org.apache.hadoop.fs.Path) SemanticException(org.apache.hadoop.hive.ql.parse.SemanticException) MapWork(org.apache.hadoop.hive.ql.plan.MapWork) OperatorDesc(org.apache.hadoop.hive.ql.plan.OperatorDesc) JoinDesc(org.apache.hadoop.hive.ql.plan.JoinDesc)

Example 24 with JoinDesc

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

the class GenMRSkewJoinProcessor method processSkewJoin.

/**
 * Create tasks for processing skew joins. The idea is (HIVE-964) to use
 * separated jobs and map-joins to handle skew joins.
 * <p>
 * <ul>
 * <li>
 * Number of mr jobs to handle skew keys is the number of table minus 1 (we
 * can stream the last table, so big keys in the last table will not be a
 * problem).
 * <li>
 * At runtime in Join, we output big keys in one table into one corresponding
 * directories, and all same keys in other tables into different dirs(one for
 * each table). The directories will look like:
 * <ul>
 * <li>
 * dir-T1-bigkeys(containing big keys in T1), dir-T2-keys(containing keys
 * which is big in T1),dir-T3-keys(containing keys which is big in T1), ...
 * <li>
 * dir-T1-keys(containing keys which is big in T2), dir-T2-bigkeys(containing
 * big keys in T2),dir-T3-keys(containing keys which is big in T2), ...
 * <li>
 * dir-T1-keys(containing keys which is big in T3), dir-T2-keys(containing big
 * keys in T3),dir-T3-bigkeys(containing keys which is big in T3), ... .....
 * </ul>
 * </ul>
 * For each table, we launch one mapjoin job, taking the directory containing
 * big keys in this table and corresponding dirs in other tables as input.
 * (Actually one job for one row in the above.)
 *
 * <p>
 * For more discussions, please check
 * https://issues.apache.org/jira/browse/HIVE-964.
 */
@SuppressWarnings("unchecked")
public static void processSkewJoin(JoinOperator joinOp, Task<?> currTask, ParseContext parseCtx) throws SemanticException {
    // now does not work with outer joins
    if (!GenMRSkewJoinProcessor.skewJoinEnabled(parseCtx.getConf(), joinOp)) {
        return;
    }
    List<Task<?>> children = currTask.getChildTasks();
    Path baseTmpDir = parseCtx.getContext().getMRTmpPath();
    JoinDesc joinDescriptor = joinOp.getConf();
    Map<Byte, List<ExprNodeDesc>> joinValues = joinDescriptor.getExprs();
    int numAliases = joinValues.size();
    Map<Byte, Path> bigKeysDirMap = new HashMap<Byte, Path>();
    Map<Byte, Map<Byte, Path>> smallKeysDirMap = new HashMap<Byte, Map<Byte, Path>>();
    Map<Byte, Path> skewJoinJobResultsDir = new HashMap<Byte, Path>();
    Byte[] tags = joinDescriptor.getTagOrder();
    for (int i = 0; i < numAliases; i++) {
        Byte alias = tags[i];
        bigKeysDirMap.put(alias, getBigKeysDir(baseTmpDir, alias));
        Map<Byte, Path> smallKeysMap = new HashMap<Byte, Path>();
        smallKeysDirMap.put(alias, smallKeysMap);
        for (Byte src2 : tags) {
            if (!src2.equals(alias)) {
                smallKeysMap.put(src2, getSmallKeysDir(baseTmpDir, alias, src2));
            }
        }
        skewJoinJobResultsDir.put(alias, getBigKeysSkewJoinResultDir(baseTmpDir, alias));
    }
    joinDescriptor.setHandleSkewJoin(true);
    joinDescriptor.setBigKeysDirMap(bigKeysDirMap);
    joinDescriptor.setSmallKeysDirMap(smallKeysDirMap);
    joinDescriptor.setSkewKeyDefinition(HiveConf.getIntVar(parseCtx.getConf(), HiveConf.ConfVars.HIVESKEWJOINKEY));
    HashMap<Path, Task<?>> bigKeysDirToTaskMap = new HashMap<Path, Task<?>>();
    List<Serializable> listWorks = new ArrayList<Serializable>();
    List<Task<?>> listTasks = new ArrayList<Task<?>>();
    MapredWork currPlan = (MapredWork) currTask.getWork();
    TableDesc keyTblDesc = (TableDesc) currPlan.getReduceWork().getKeyDesc().clone();
    List<String> joinKeys = Utilities.getColumnNames(keyTblDesc.getProperties());
    List<String> joinKeyTypes = Utilities.getColumnTypes(keyTblDesc.getProperties());
    Map<Byte, TableDesc> tableDescList = new HashMap<Byte, TableDesc>();
    Map<Byte, RowSchema> rowSchemaList = new HashMap<Byte, RowSchema>();
    Map<Byte, List<ExprNodeDesc>> newJoinValues = new HashMap<Byte, List<ExprNodeDesc>>();
    Map<Byte, List<ExprNodeDesc>> newJoinKeys = new HashMap<Byte, List<ExprNodeDesc>>();
    // used for create mapJoinDesc, should be in order
    List<TableDesc> newJoinValueTblDesc = new ArrayList<TableDesc>();
    for (Byte tag : tags) {
        newJoinValueTblDesc.add(null);
    }
    for (int i = 0; i < numAliases; i++) {
        Byte alias = tags[i];
        List<ExprNodeDesc> valueCols = joinValues.get(alias);
        String colNames = "";
        String colTypes = "";
        int columnSize = valueCols.size();
        List<ExprNodeDesc> newValueExpr = new ArrayList<ExprNodeDesc>();
        List<ExprNodeDesc> newKeyExpr = new ArrayList<ExprNodeDesc>();
        ArrayList<ColumnInfo> columnInfos = new ArrayList<ColumnInfo>();
        boolean first = true;
        for (int k = 0; k < columnSize; k++) {
            TypeInfo type = valueCols.get(k).getTypeInfo();
            // any name, it does not matter.
            String newColName = i + "_VALUE_" + k;
            ColumnInfo columnInfo = new ColumnInfo(newColName, type, alias.toString(), false);
            columnInfos.add(columnInfo);
            newValueExpr.add(new ExprNodeColumnDesc(columnInfo));
            if (!first) {
                colNames = colNames + ",";
                colTypes = colTypes + ",";
            }
            first = false;
            colNames = colNames + newColName;
            colTypes = colTypes + valueCols.get(k).getTypeString();
        }
        // we are putting join keys at last part of the spilled table
        for (int k = 0; k < joinKeys.size(); k++) {
            if (!first) {
                colNames = colNames + ",";
                colTypes = colTypes + ",";
            }
            first = false;
            colNames = colNames + joinKeys.get(k);
            colTypes = colTypes + joinKeyTypes.get(k);
            ColumnInfo columnInfo = new ColumnInfo(joinKeys.get(k), TypeInfoFactory.getPrimitiveTypeInfo(joinKeyTypes.get(k)), alias.toString(), false);
            columnInfos.add(columnInfo);
            newKeyExpr.add(new ExprNodeColumnDesc(columnInfo));
        }
        newJoinValues.put(alias, newValueExpr);
        newJoinKeys.put(alias, newKeyExpr);
        tableDescList.put(alias, Utilities.getTableDesc(colNames, colTypes));
        rowSchemaList.put(alias, new RowSchema(columnInfos));
        // construct value table Desc
        String valueColNames = "";
        String valueColTypes = "";
        first = true;
        for (int k = 0; k < columnSize; k++) {
            // any name, it does not matter.
            String newColName = i + "_VALUE_" + k;
            if (!first) {
                valueColNames = valueColNames + ",";
                valueColTypes = valueColTypes + ",";
            }
            valueColNames = valueColNames + newColName;
            valueColTypes = valueColTypes + valueCols.get(k).getTypeString();
            first = false;
        }
        newJoinValueTblDesc.set(Byte.valueOf((byte) i), Utilities.getTableDesc(valueColNames, valueColTypes));
    }
    joinDescriptor.setSkewKeysValuesTables(tableDescList);
    joinDescriptor.setKeyTableDesc(keyTblDesc);
    for (int i = 0; i < numAliases - 1; i++) {
        Byte src = tags[i];
        MapWork newPlan = PlanUtils.getMapRedWork().getMapWork();
        // This code has been only added for testing
        boolean mapperCannotSpanPartns = parseCtx.getConf().getBoolVar(HiveConf.ConfVars.HIVE_MAPPER_CANNOT_SPAN_MULTIPLE_PARTITIONS);
        newPlan.setMapperCannotSpanPartns(mapperCannotSpanPartns);
        MapredWork clonePlan = SerializationUtilities.clonePlan(currPlan);
        Operator<? extends OperatorDesc>[] parentOps = new TableScanOperator[tags.length];
        for (int k = 0; k < tags.length; k++) {
            Operator<? extends OperatorDesc> ts = GenMapRedUtils.createTemporaryTableScanOperator(joinOp.getCompilationOpContext(), rowSchemaList.get((byte) k));
            ((TableScanOperator) ts).setTableDescSkewJoin(tableDescList.get((byte) k));
            parentOps[k] = ts;
        }
        Operator<? extends OperatorDesc> tblScan_op = parentOps[i];
        ArrayList<String> aliases = new ArrayList<String>();
        String alias = src.toString().intern();
        aliases.add(alias);
        Path bigKeyDirPath = bigKeysDirMap.get(src);
        newPlan.addPathToAlias(bigKeyDirPath, aliases);
        newPlan.getAliasToWork().put(alias, tblScan_op);
        PartitionDesc part = new PartitionDesc(tableDescList.get(src), null);
        newPlan.addPathToPartitionInfo(bigKeyDirPath, part);
        newPlan.getAliasToPartnInfo().put(alias, part);
        Operator<? extends OperatorDesc> reducer = clonePlan.getReduceWork().getReducer();
        assert reducer instanceof JoinOperator;
        JoinOperator cloneJoinOp = (JoinOperator) reducer;
        String dumpFilePrefix = "mapfile" + PlanUtils.getCountForMapJoinDumpFilePrefix();
        MapJoinDesc mapJoinDescriptor = new MapJoinDesc(newJoinKeys, keyTblDesc, newJoinValues, newJoinValueTblDesc, newJoinValueTblDesc, joinDescriptor.getOutputColumnNames(), i, joinDescriptor.getConds(), joinDescriptor.getFilters(), joinDescriptor.getNoOuterJoin(), dumpFilePrefix, joinDescriptor.getMemoryMonitorInfo(), joinDescriptor.getInMemoryDataSize());
        mapJoinDescriptor.setTagOrder(tags);
        mapJoinDescriptor.setHandleSkewJoin(false);
        mapJoinDescriptor.setNullSafes(joinDescriptor.getNullSafes());
        mapJoinDescriptor.setColumnExprMap(joinDescriptor.getColumnExprMap());
        MapredLocalWork localPlan = new MapredLocalWork(new LinkedHashMap<String, Operator<? extends OperatorDesc>>(), new LinkedHashMap<String, FetchWork>());
        Map<Byte, Path> smallTblDirs = smallKeysDirMap.get(src);
        for (int j = 0; j < numAliases; j++) {
            if (j == i) {
                continue;
            }
            Byte small_alias = tags[j];
            Operator<? extends OperatorDesc> tblScan_op2 = parentOps[j];
            localPlan.getAliasToWork().put(small_alias.toString(), tblScan_op2);
            Path tblDir = smallTblDirs.get(small_alias);
            localPlan.getAliasToFetchWork().put(small_alias.toString(), new FetchWork(tblDir, tableDescList.get(small_alias)));
        }
        newPlan.setMapRedLocalWork(localPlan);
        // construct a map join and set it as the child operator of tblScan_op
        MapJoinOperator mapJoinOp = (MapJoinOperator) OperatorFactory.getAndMakeChild(joinOp.getCompilationOpContext(), mapJoinDescriptor, (RowSchema) null, parentOps);
        // change the children of the original join operator to point to the map
        // join operator
        List<Operator<? extends OperatorDesc>> childOps = cloneJoinOp.getChildOperators();
        for (Operator<? extends OperatorDesc> childOp : childOps) {
            childOp.replaceParent(cloneJoinOp, mapJoinOp);
        }
        mapJoinOp.setChildOperators(childOps);
        HiveConf jc = new HiveConf(parseCtx.getConf(), GenMRSkewJoinProcessor.class);
        newPlan.setNumMapTasks(HiveConf.getIntVar(jc, HiveConf.ConfVars.HIVESKEWJOINMAPJOINNUMMAPTASK));
        newPlan.setMinSplitSize(HiveConf.getLongVar(jc, HiveConf.ConfVars.HIVESKEWJOINMAPJOINMINSPLIT));
        newPlan.setInputformat(HiveInputFormat.class.getName());
        MapredWork w = new MapredWork();
        w.setMapWork(newPlan);
        Task<?> skewJoinMapJoinTask = TaskFactory.get(w);
        skewJoinMapJoinTask.setFetchSource(currTask.isFetchSource());
        bigKeysDirToTaskMap.put(bigKeyDirPath, skewJoinMapJoinTask);
        listWorks.add(skewJoinMapJoinTask.getWork());
        listTasks.add(skewJoinMapJoinTask);
    }
    if (children != null) {
        for (Task<?> tsk : listTasks) {
            for (Task<?> oldChild : children) {
                tsk.addDependentTask(oldChild);
            }
        }
        currTask.setChildTasks(new ArrayList<Task<?>>());
        for (Task<?> oldChild : children) {
            oldChild.getParentTasks().remove(currTask);
        }
        listTasks.addAll(children);
    }
    ConditionalResolverSkewJoinCtx context = new ConditionalResolverSkewJoinCtx(bigKeysDirToTaskMap, children);
    ConditionalWork cndWork = new ConditionalWork(listWorks);
    ConditionalTask cndTsk = (ConditionalTask) TaskFactory.get(cndWork);
    cndTsk.setListTasks(listTasks);
    cndTsk.setResolver(new ConditionalResolverSkewJoin());
    cndTsk.setResolverCtx(context);
    currTask.setChildTasks(new ArrayList<Task<?>>());
    currTask.addDependentTask(cndTsk);
    return;
}
Also used : MapJoinOperator(org.apache.hadoop.hive.ql.exec.MapJoinOperator) JoinOperator(org.apache.hadoop.hive.ql.exec.JoinOperator) ConditionalTask(org.apache.hadoop.hive.ql.exec.ConditionalTask) Task(org.apache.hadoop.hive.ql.exec.Task) Serializable(java.io.Serializable) TableScanOperator(org.apache.hadoop.hive.ql.exec.TableScanOperator) HashMap(java.util.HashMap) LinkedHashMap(java.util.LinkedHashMap) ArrayList(java.util.ArrayList) ConditionalWork(org.apache.hadoop.hive.ql.plan.ConditionalWork) ColumnInfo(org.apache.hadoop.hive.ql.exec.ColumnInfo) MapredWork(org.apache.hadoop.hive.ql.plan.MapredWork) ConditionalTask(org.apache.hadoop.hive.ql.exec.ConditionalTask) ArrayList(java.util.ArrayList) List(java.util.List) HiveConf(org.apache.hadoop.hive.conf.HiveConf) ExprNodeDesc(org.apache.hadoop.hive.ql.plan.ExprNodeDesc) ConditionalResolverSkewJoin(org.apache.hadoop.hive.ql.plan.ConditionalResolverSkewJoin) MapJoinOperator(org.apache.hadoop.hive.ql.exec.MapJoinOperator) RowSchema(org.apache.hadoop.hive.ql.exec.RowSchema) TypeInfo(org.apache.hadoop.hive.serde2.typeinfo.TypeInfo) MapredLocalWork(org.apache.hadoop.hive.ql.plan.MapredLocalWork) TableDesc(org.apache.hadoop.hive.ql.plan.TableDesc) MapJoinDesc(org.apache.hadoop.hive.ql.plan.MapJoinDesc) JoinDesc(org.apache.hadoop.hive.ql.plan.JoinDesc) HashMap(java.util.HashMap) LinkedHashMap(java.util.LinkedHashMap) Map(java.util.Map) OperatorDesc(org.apache.hadoop.hive.ql.plan.OperatorDesc) MapJoinOperator(org.apache.hadoop.hive.ql.exec.MapJoinOperator) JoinOperator(org.apache.hadoop.hive.ql.exec.JoinOperator) TableScanOperator(org.apache.hadoop.hive.ql.exec.TableScanOperator) Operator(org.apache.hadoop.hive.ql.exec.Operator) HiveInputFormat(org.apache.hadoop.hive.ql.io.HiveInputFormat) ExprNodeColumnDesc(org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc) Path(org.apache.hadoop.fs.Path) MapJoinDesc(org.apache.hadoop.hive.ql.plan.MapJoinDesc) ConditionalResolverSkewJoinCtx(org.apache.hadoop.hive.ql.plan.ConditionalResolverSkewJoin.ConditionalResolverSkewJoinCtx) MapWork(org.apache.hadoop.hive.ql.plan.MapWork) FetchWork(org.apache.hadoop.hive.ql.plan.FetchWork) PartitionDesc(org.apache.hadoop.hive.ql.plan.PartitionDesc)

Aggregations

JoinDesc (org.apache.hadoop.hive.ql.plan.JoinDesc)24 ArrayList (java.util.ArrayList)15 MapJoinDesc (org.apache.hadoop.hive.ql.plan.MapJoinDesc)15 Operator (org.apache.hadoop.hive.ql.exec.Operator)14 JoinOperator (org.apache.hadoop.hive.ql.exec.JoinOperator)13 List (java.util.List)12 TableScanOperator (org.apache.hadoop.hive.ql.exec.TableScanOperator)12 JoinCondDesc (org.apache.hadoop.hive.ql.plan.JoinCondDesc)12 HashMap (java.util.HashMap)11 ExprNodeDesc (org.apache.hadoop.hive.ql.plan.ExprNodeDesc)11 ReduceSinkOperator (org.apache.hadoop.hive.ql.exec.ReduceSinkOperator)10 OperatorDesc (org.apache.hadoop.hive.ql.plan.OperatorDesc)10 LinkedHashMap (java.util.LinkedHashMap)9 ColumnInfo (org.apache.hadoop.hive.ql.exec.ColumnInfo)9 SemanticException (org.apache.hadoop.hive.ql.parse.SemanticException)9 MapJoinOperator (org.apache.hadoop.hive.ql.exec.MapJoinOperator)8 RowSchema (org.apache.hadoop.hive.ql.exec.RowSchema)8 ExprNodeColumnDesc (org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc)8 SelectOperator (org.apache.hadoop.hive.ql.exec.SelectOperator)7 HashSet (java.util.HashSet)6