Examples with TezTask org.apache.hadoop.hive.ql.exec.tez.TezTask used on opensource projects

Example 1 with TezTask

use of org.apache.hadoop.hive.ql.exec.tez.TezTask in project hive by apache.

the class PostExecTezSummaryPrinter method run.

@Override
public void run(HookContext hookContext) throws Exception {
    assert (hookContext.getHookType() == HookContext.HookType.POST_EXEC_HOOK);
    HiveConf conf = hookContext.getConf();
    if (!"tez".equals(HiveConf.getVar(conf, HiveConf.ConfVars.HIVE_EXECUTION_ENGINE))) {
        return;
    }
    LOG.info("Executing post execution hook to print tez summary..");
    SessionState ss = SessionState.get();
    SessionState.LogHelper console = ss.getConsole();
    QueryPlan plan = hookContext.getQueryPlan();
    if (plan == null) {
        return;
    }
    List<TezTask> rootTasks = Utilities.getTezTasks(plan.getRootTasks());
    for (TezTask tezTask : rootTasks) {
        LOG.info("Printing summary for tez task: " + tezTask.getName());
        TezCounters counters = tezTask.getTezCounters();
        if (counters != null) {
            String hiveCountersGroup = HiveConf.getVar(conf, HiveConf.ConfVars.HIVECOUNTERGROUP);
            for (CounterGroup group : counters) {
                if (hiveCountersGroup.equals(group.getDisplayName())) {
                    console.printError(tezTask.getId() + " HIVE COUNTERS:");
                    for (TezCounter counter : group) {
                        console.printError("   " + counter.getDisplayName() + ": " + counter.getValue());
                    }
                } else if (group.getName().equals(FileSystemCounter.class.getName())) {
                    console.printError(tezTask.getId() + " FILE SYSTEM COUNTERS:");
                    for (TezCounter counter : group) {
                        // local file system counters
                        if (counter.getName().contains("HDFS")) {
                            console.printError("   " + counter.getDisplayName() + ": " + counter.getValue());
                        }
                    }
                } else if (group.getName().equals(LlapIOCounters.class.getName())) {
                    console.printError(tezTask.getId() + " LLAP IO COUNTERS:");
                    List<String> testSafeCounters = LlapIOCounters.testSafeCounterNames();
                    for (TezCounter counter : group) {
                        if (testSafeCounters.contains(counter.getDisplayName())) {
                            console.printError("   " + counter.getDisplayName() + ": " + counter.getValue());
                        }
                    }
                }
            }
        }
    }
}

Example 2 with TezTask

use of org.apache.hadoop.hive.ql.exec.tez.TezTask in project hive by apache.

the class ATSHook method determineLlapId.

private ApplicationId determineLlapId(final HiveConf conf, QueryPlan plan) throws IOException {
    //       be added here, although this is only necessary to have extra debug information.
    for (TezTask tezTask : Utilities.getTezTasks(plan.getRootTasks())) {
        if (!tezTask.getWork().getLlapMode())
            continue;
        // In HS2, the client should have been cached already for the common case.
        // Otherwise, this may actually introduce delay to compilation for the first query.
        String hosts = HiveConf.getVar(conf, HiveConf.ConfVars.LLAP_DAEMON_SERVICE_HOSTS);
        if (hosts != null && !hosts.isEmpty()) {
            ApplicationId llapId = LlapRegistryService.getClient(conf).getApplicationId();
            LOG.info("The query will use LLAP instance " + llapId + " (" + hosts + ")");
            return llapId;
        } else {
            LOG.info("Cannot determine LLAP instance on client - service hosts are not set");
            return null;
        }
    }
    return null;
}

Example 3 with TezTask

use of org.apache.hadoop.hive.ql.exec.tez.TezTask in project hive by apache.

the class CrossProductCheck method dispatch.

@Override
public Object dispatch(Node nd, Stack<Node> stack, Object... nodeOutputs) throws SemanticException {
    @SuppressWarnings("unchecked") Task<? extends Serializable> currTask = (Task<? extends Serializable>) nd;
    if (currTask instanceof MapRedTask) {
        MapRedTask mrTsk = (MapRedTask) currTask;
        MapredWork mrWrk = mrTsk.getWork();
        checkMapJoins(mrTsk);
        checkMRReducer(currTask.toString(), mrWrk);
    } else if (currTask instanceof ConditionalTask) {
        List<Task<? extends Serializable>> taskListInConditionalTask = ((ConditionalTask) currTask).getListTasks();
        for (Task<? extends Serializable> tsk : taskListInConditionalTask) {
            dispatch(tsk, stack, nodeOutputs);
        }
    } else if (currTask instanceof TezTask) {
        TezTask tzTask = (TezTask) currTask;
        TezWork tzWrk = tzTask.getWork();
        checkMapJoins(tzWrk);
        checkTezReducer(tzWrk);
    }
    return null;
}

Example 4 with TezTask

use of org.apache.hadoop.hive.ql.exec.tez.TezTask in project hive by apache.

the class GenericUDTFGetSplits method createPlanFragment.

public PlanFragment createPlanFragment(String query, int num) throws HiveException {
    HiveConf conf = new HiveConf(SessionState.get().getConf());
    HiveConf.setVar(conf, ConfVars.HIVEFETCHTASKCONVERSION, "none");
    HiveConf.setVar(conf, ConfVars.HIVEQUERYRESULTFILEFORMAT, PlanUtils.LLAP_OUTPUT_FORMAT_KEY);
    String originalMode = HiveConf.getVar(conf, ConfVars.HIVE_EXECUTION_MODE);
    HiveConf.setVar(conf, ConfVars.HIVE_EXECUTION_MODE, "llap");
    HiveConf.setBoolVar(conf, ConfVars.HIVE_TEZ_GENERATE_CONSISTENT_SPLITS, true);
    HiveConf.setBoolVar(conf, ConfVars.LLAP_CLIENT_CONSISTENT_SPLITS, true);
    conf.setBoolean(TezSplitGrouper.TEZ_GROUPING_NODE_LOCAL_ONLY, true);
    // Tez/LLAP requires RPC query plan
    HiveConf.setBoolVar(conf, ConfVars.HIVE_RPC_QUERY_PLAN, true);
    try {
        jc = DagUtils.getInstance().createConfiguration(conf);
    } catch (IOException e) {
        throw new HiveException(e);
    }
    Driver driver = new Driver(conf);
    try {
        CommandProcessorResponse cpr = driver.compileAndRespond(query);
        if (cpr.getResponseCode() != 0) {
            throw new HiveException("Failed to compile query: " + cpr.getException());
        }
        QueryPlan plan = driver.getPlan();
        List<Task<?>> roots = plan.getRootTasks();
        Schema schema = convertSchema(plan.getResultSchema());
        if (roots == null || roots.size() != 1 || !(roots.get(0) instanceof TezTask)) {
            throw new HiveException("Was expecting a single TezTask.");
        }
        TezWork tezWork = ((TezTask) roots.get(0)).getWork();
        if (tezWork.getAllWork().size() != 1) {
            String tableName = "table_" + UUID.randomUUID().toString().replaceAll("[^A-Za-z0-9 ]", "");
            String ctas = "create temporary table " + tableName + " as " + query;
            LOG.info("Materializing the query for LLAPIF; CTAS: " + ctas);
            try {
                driver.resetQueryState();
                HiveConf.setVar(conf, ConfVars.HIVE_EXECUTION_MODE, originalMode);
                cpr = driver.run(ctas, false);
            } catch (CommandNeedRetryException e) {
                throw new HiveException(e);
            }
            if (cpr.getResponseCode() != 0) {
                throw new HiveException("Failed to create temp table: " + cpr.getException());
            }
            HiveConf.setVar(conf, ConfVars.HIVE_EXECUTION_MODE, "llap");
            query = "select * from " + tableName;
            cpr = driver.compileAndRespond(query);
            if (cpr.getResponseCode() != 0) {
                throw new HiveException("Failed to create temp table: " + cpr.getException());
            }
            plan = driver.getPlan();
            roots = plan.getRootTasks();
            schema = convertSchema(plan.getResultSchema());
            if (roots == null || roots.size() != 1 || !(roots.get(0) instanceof TezTask)) {
                throw new HiveException("Was expecting a single TezTask.");
            }
            tezWork = ((TezTask) roots.get(0)).getWork();
        }
        return new PlanFragment(tezWork, schema, jc);
    } finally {
        driver.close();
        driver.destroy();
    }
}

Example 5 with TezTask

use of org.apache.hadoop.hive.ql.exec.tez.TezTask in project hive by apache.

the class PostExecOrcRowGroupCountPrinter method run.

@Override
public void run(HookContext hookContext) throws Exception {
    assert (hookContext.getHookType() == HookContext.HookType.POST_EXEC_HOOK);
    HiveConf conf = hookContext.getConf();
    if (!"tez".equals(HiveConf.getVar(conf, HiveConf.ConfVars.HIVE_EXECUTION_ENGINE))) {
        return;
    }
    LOG.info("Executing post execution hook to print ORC row groups read counter..");
    SessionState ss = SessionState.get();
    SessionState.LogHelper console = ss.getConsole();
    QueryPlan plan = hookContext.getQueryPlan();
    if (plan == null) {
        return;
    }
    List<TezTask> rootTasks = Utilities.getTezTasks(plan.getRootTasks());
    for (TezTask tezTask : rootTasks) {
        LOG.info("Printing ORC row group counter for tez task: " + tezTask.getName());
        TezCounters counters = tezTask.getTezCounters();
        if (counters != null) {
            for (CounterGroup group : counters) {
                if (group.getName().equals(LlapIOCounters.class.getName())) {
                    console.printError(tezTask.getId() + " LLAP IO COUNTERS:");
                    for (TezCounter counter : group) {
                        if (counter.getDisplayName().equals(LlapIOCounters.SELECTED_ROWGROUPS.name())) {
                            console.printError("   " + counter.getDisplayName() + ": " + counter.getValue());
                        }
                    }
                }
            }
        }
    }
}