org.apache.hadoop.hive.ql.exec.mr.HadoopJobExecHelper.java Source code

Java tutorial

Introduction

Here is the source code for org.apache.hadoop.hive.ql.exec.mr.HadoopJobExecHelper.java

Source

/**
 * Licensed to the Apache Software Foundation (ASF) under one
 * or more contributor license agreements.  See the NOTICE file
 * distributed with this work for additional information
 * regarding copyright ownership.  The ASF licenses this file
 * to you under the Apache License, Version 2.0 (the
 * "License"); you may not use this file except in compliance
 * with the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package org.apache.hadoop.hive.ql.exec.mr;

import java.io.IOException;
import java.io.Serializable;
import java.text.SimpleDateFormat;
import java.util.ArrayList;
import java.util.Calendar;
import java.util.Collections;
import java.util.Enumeration;
import java.util.HashMap;
import java.util.LinkedList;
import java.util.List;
import java.util.Map;
import java.util.concurrent.TimeUnit;

import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hive.common.JavaUtils;
import org.apache.hadoop.hive.conf.HiveConf;
import org.apache.hadoop.hive.conf.HiveConf.ConfVars;
import org.apache.hadoop.hive.ql.MapRedStats;
import org.apache.hadoop.hive.ql.exec.Heartbeater;
import org.apache.hadoop.hive.ql.exec.Operator;
import org.apache.hadoop.hive.ql.exec.Task;
import org.apache.hadoop.hive.ql.exec.TaskHandle;
import org.apache.hadoop.hive.ql.exec.Utilities;
import org.apache.hadoop.hive.ql.history.HiveHistory.Keys;
import org.apache.hadoop.hive.ql.lockmgr.HiveTxnManager;
import org.apache.hadoop.hive.ql.plan.ReducerTimeStatsPerJob;
import org.apache.hadoop.hive.ql.session.SessionState;
import org.apache.hadoop.hive.ql.session.SessionState.LogHelper;
import org.apache.hadoop.hive.ql.stats.ClientStatsPublisher;
import org.apache.hadoop.hive.shims.ShimLoader;
import org.apache.hadoop.mapred.Counters;
import org.apache.hadoop.mapred.Counters.Counter;
import org.apache.hadoop.mapred.JobClient;
import org.apache.hadoop.mapred.JobConf;
import org.apache.hadoop.mapred.JobID;
import org.apache.hadoop.mapred.JobStatus;
import org.apache.hadoop.mapred.RunningJob;
import org.apache.hadoop.mapred.TaskCompletionEvent;
import org.apache.hadoop.mapred.TaskReport;
import org.apache.log4j.Appender;
import org.apache.log4j.FileAppender;
import org.apache.log4j.LogManager;

public class HadoopJobExecHelper {

    static final private Log LOG = LogFactory.getLog(HadoopJobExecHelper.class.getName());

    protected transient JobConf job;
    protected Task<? extends Serializable> task;

    protected transient int mapProgress = -1;
    protected transient int reduceProgress = -1;

    protected transient int lastMapProgress;
    protected transient int lastReduceProgress;

    public transient JobID jobId;
    private LogHelper console;
    private HadoopJobExecHook callBackObj;

    /**
     * Update counters relevant to this task.
     */
    private void updateCounters(Counters ctrs, RunningJob rj) throws IOException {
        lastMapProgress = mapProgress;
        lastReduceProgress = reduceProgress;
        mapProgress = Math.round(rj.mapProgress() * 100);
        mapProgress = mapProgress == 100 ? (int) Math.floor(rj.mapProgress() * 100) : mapProgress;
        reduceProgress = Math.round(rj.reduceProgress() * 100);
        reduceProgress = reduceProgress == 100 ? (int) Math.floor(rj.reduceProgress() * 100) : reduceProgress;
        task.taskCounters.put("CNTR_NAME_" + task.getId() + "_MAP_PROGRESS", Long.valueOf(mapProgress));
        task.taskCounters.put("CNTR_NAME_" + task.getId() + "_REDUCE_PROGRESS", Long.valueOf(reduceProgress));
    }

    /**
     * This msg pattern is used to track when a job is started.
     *
     * @param jobId
     * @return
     */
    private static String getJobStartMsg(JobID jobId) {
        return "Starting Job = " + jobId;
    }

    /**
     * this msg pattern is used to track when a job is successfully done.
     *
     * @param jobId
     * @return the job end message
     */
    public static String getJobEndMsg(JobID jobId) {
        return "Ended Job = " + jobId;
    }

    public boolean mapStarted() {
        return mapProgress > 0;
    }

    public boolean reduceStarted() {
        return reduceProgress > 0;
    }

    public boolean mapDone() {
        return mapProgress == 100;
    }

    public boolean reduceDone() {
        return reduceProgress == 100;
    }

    public JobID getJobId() {
        return jobId;
    }

    public void setJobId(JobID jobId) {
        this.jobId = jobId;
    }

    public HadoopJobExecHelper(JobConf job, LogHelper console, Task<? extends Serializable> task,
            HadoopJobExecHook hookCallBack) {
        this.job = job;
        this.console = console;
        this.task = task;
        this.callBackObj = hookCallBack;

        if (job != null) {
            // even with tez on some jobs are run as MR. disable the flag in
            // the conf, so that the backend runs fully as MR.
            HiveConf.setVar(job, HiveConf.ConfVars.HIVE_EXECUTION_ENGINE, "mr");
        }
    }

    /**
     * A list of the currently running jobs spawned in this Hive instance that is used to kill all
     * running jobs in the event of an unexpected shutdown - i.e., the JVM shuts down while there are
     * still jobs running.
     */
    public static List<RunningJob> runningJobs = Collections.synchronizedList(new LinkedList<RunningJob>());

    /**
     * In Hive, when the user control-c's the command line, any running jobs spawned from that command
     * line are best-effort killed.
     *
     * This static constructor registers a shutdown thread to iterate over all the running job kill
     * URLs and do a get on them.
     *
     */
    static {
        Runtime.getRuntime().addShutdownHook(new Thread() {
            @Override
            public void run() {
                killRunningJobs();
            }
        });
    }

    public static void killRunningJobs() {
        synchronized (runningJobs) {
            for (RunningJob rj : runningJobs) {
                try {
                    System.err.println("killing job with: " + rj.getID());
                    rj.killJob();
                } catch (Exception e) {
                    LOG.warn(e);
                    System.err.println("Failed to kill job: " + rj.getID());
                    // do nothing
                }
            }
        }
    }

    @SuppressWarnings("deprecation")
    public boolean checkFatalErrors(Counters ctrs, StringBuilder errMsg) {
        if (ctrs == null) {
            // hadoop might return null if it cannot locate the job.
            // we may still be able to retrieve the job status - so ignore
            return false;
        }
        // check for number of created files
        Counters.Counter cntr = ctrs.findCounter(HiveConf.getVar(job, ConfVars.HIVECOUNTERGROUP),
                Operator.HIVECOUNTERCREATEDFILES);
        long numFiles = cntr != null ? cntr.getValue() : 0;
        long upperLimit = HiveConf.getLongVar(job, HiveConf.ConfVars.MAXCREATEDFILES);
        if (numFiles > upperLimit) {
            errMsg.append("total number of created files now is " + numFiles + ", which exceeds ")
                    .append(upperLimit);
            return true;
        }
        return this.callBackObj.checkFatalErrors(ctrs, errMsg);
    }

    private MapRedStats progress(ExecDriverTaskHandle th) throws IOException {
        JobClient jc = th.getJobClient();
        RunningJob rj = th.getRunningJob();
        SimpleDateFormat dateFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss,SSS");
        //DecimalFormat longFormatter = new DecimalFormat("###,###");
        long reportTime = System.currentTimeMillis();
        long maxReportInterval = HiveConf.getTimeVar(job,
                HiveConf.ConfVars.HIVE_LOG_INCREMENTAL_PLAN_PROGRESS_INTERVAL, TimeUnit.MILLISECONDS);
        boolean fatal = false;
        StringBuilder errMsg = new StringBuilder();
        long pullInterval = HiveConf.getLongVar(job, HiveConf.ConfVars.HIVECOUNTERSPULLINTERVAL);
        boolean initializing = true;
        boolean initOutputPrinted = false;
        long cpuMsec = -1;
        int numMap = -1;
        int numReduce = -1;
        List<ClientStatsPublisher> clientStatPublishers = getClientStatPublishers();
        final boolean localMode = ShimLoader.getHadoopShims().isLocalMode(job);
        Heartbeater heartbeater = new Heartbeater(th.getTxnManager(), job);

        while (!rj.isComplete()) {
            try {
                Thread.sleep(pullInterval);
            } catch (InterruptedException e) {
            }
            heartbeater.heartbeat();

            if (initializing && rj.getJobState() == JobStatus.PREP) {
                // No reason to poll untill the job is initialized
                continue;
            } else {
                // By now the job is initialized so no reason to do
                // rj.getJobState() again and we do not want to do an extra RPC call
                initializing = false;
            }

            if (!localMode) {
                if (!initOutputPrinted) {
                    SessionState ss = SessionState.get();

                    String logMapper;
                    String logReducer;

                    TaskReport[] mappers = jc.getMapTaskReports(rj.getID());
                    if (mappers == null) {
                        logMapper = "no information for number of mappers; ";
                    } else {
                        numMap = mappers.length;
                        if (ss != null) {
                            ss.getHiveHistory().setTaskProperty(SessionState.get().getQueryId(), getId(),
                                    Keys.TASK_NUM_MAPPERS, Integer.toString(numMap));
                        }
                        logMapper = "number of mappers: " + numMap + "; ";
                    }

                    TaskReport[] reducers = jc.getReduceTaskReports(rj.getID());
                    if (reducers == null) {
                        logReducer = "no information for number of reducers. ";
                    } else {
                        numReduce = reducers.length;
                        if (ss != null) {
                            ss.getHiveHistory().setTaskProperty(SessionState.get().getQueryId(), getId(),
                                    Keys.TASK_NUM_REDUCERS, Integer.toString(numReduce));
                        }
                        logReducer = "number of reducers: " + numReduce;
                    }

                    console.printInfo("Hadoop job information for " + getId() + ": " + logMapper + logReducer);
                    initOutputPrinted = true;
                }

                RunningJob newRj = jc.getJob(rj.getID());
                if (newRj == null) {
                    // under exceptional load, hadoop may not be able to look up status
                    // of finished jobs (because it has purged them from memory). From
                    // hive's perspective - it's equivalent to the job having failed.
                    // So raise a meaningful exception
                    throw new IOException("Could not find status of job:" + rj.getID());
                } else {
                    th.setRunningJob(newRj);
                    rj = newRj;
                }
            }

            // If fatal errors happen we should kill the job immediately rather than
            // let the job retry several times, which eventually lead to failure.
            if (fatal) {
                continue; // wait until rj.isComplete
            }

            Counters ctrs = th.getCounters();

            if (fatal = checkFatalErrors(ctrs, errMsg)) {
                console.printError("[Fatal Error] " + errMsg.toString() + ". Killing the job.");
                rj.killJob();
                continue;
            }
            errMsg.setLength(0);

            updateCounters(ctrs, rj);

            // Prepare data for Client Stat Publishers (if any present) and execute them
            if (clientStatPublishers.size() > 0 && ctrs != null) {
                Map<String, Double> exctractedCounters = extractAllCounterValues(ctrs);
                for (ClientStatsPublisher clientStatPublisher : clientStatPublishers) {
                    try {
                        clientStatPublisher.run(exctractedCounters, rj.getID().toString());
                    } catch (RuntimeException runtimeException) {
                        LOG.error(
                                "Exception " + runtimeException.getClass().getCanonicalName()
                                        + " thrown when running clientStatsPublishers. The stack trace is: ",
                                runtimeException);
                    }
                }
            }

            if (mapProgress == lastMapProgress && reduceProgress == lastReduceProgress
                    && System.currentTimeMillis() < reportTime + maxReportInterval) {
                continue;
            }
            StringBuilder report = new StringBuilder();
            report.append(dateFormat.format(Calendar.getInstance().getTime()));

            report.append(' ').append(getId());
            report.append(" map = ").append(mapProgress).append("%, ");
            report.append(" reduce = ").append(reduceProgress).append('%');

            // find out CPU msecs
            // In the case that we can't find out this number, we just skip the step to print
            // it out.
            if (ctrs != null) {
                Counter counterCpuMsec = ctrs.findCounter("org.apache.hadoop.mapred.Task$Counter",
                        "CPU_MILLISECONDS");
                if (counterCpuMsec != null) {
                    long newCpuMSec = counterCpuMsec.getValue();
                    if (newCpuMSec > 0) {
                        cpuMsec = newCpuMSec;
                        report.append(", Cumulative CPU ").append((cpuMsec / 1000D)).append(" sec");
                    }
                }
            }

            // write out serialized plan with counters to log file
            // LOG.info(queryPlan);
            String output = report.toString();
            SessionState ss = SessionState.get();
            if (ss != null) {
                ss.getHiveHistory().setTaskCounters(SessionState.get().getQueryId(), getId(), ctrs);
                ss.getHiveHistory().setTaskProperty(SessionState.get().getQueryId(), getId(),
                        Keys.TASK_HADOOP_PROGRESS, output);
                if (ss.getConf().getBoolVar(HiveConf.ConfVars.HIVE_LOG_INCREMENTAL_PLAN_PROGRESS)) {
                    ss.getHiveHistory().progressTask(SessionState.get().getQueryId(), this.task);
                    this.callBackObj.logPlanProgress(ss);
                }
            }
            console.printInfo(output);
            reportTime = System.currentTimeMillis();
        }

        if (cpuMsec > 0) {
            console.printInfo("MapReduce Total cumulative CPU time: " + Utilities.formatMsecToStr(cpuMsec));
        }

        boolean success;

        Counters ctrs = th.getCounters();
        if (fatal) {
            success = false;
        } else {
            // check for fatal error again in case it occurred after
            // the last check before the job is completed
            if (checkFatalErrors(ctrs, errMsg)) {
                console.printError("[Fatal Error] " + errMsg.toString());
                success = false;
            } else {
                SessionState ss = SessionState.get();
                if (ss != null) {
                    ss.getHiveHistory().setTaskCounters(SessionState.get().getQueryId(), getId(), ctrs);
                }
                success = rj.isSuccessful();
            }
        }

        if (ctrs != null) {
            Counter counterCpuMsec = ctrs.findCounter("org.apache.hadoop.mapred.Task$Counter", "CPU_MILLISECONDS");
            if (counterCpuMsec != null) {
                long newCpuMSec = counterCpuMsec.getValue();
                if (newCpuMSec > cpuMsec) {
                    cpuMsec = newCpuMSec;
                }
            }
        }

        MapRedStats mapRedStats = new MapRedStats(numMap, numReduce, cpuMsec, success, rj.getID().toString());
        mapRedStats.setCounters(ctrs);

        // update based on the final value of the counters
        updateCounters(ctrs, rj);

        SessionState ss = SessionState.get();
        if (ss != null) {
            this.callBackObj.logPlanProgress(ss);
        }
        // LOG.info(queryPlan);
        return mapRedStats;
    }

    private String getId() {
        return this.task.getId();
    }

    /**
     * from StreamJob.java.
     */
    public void jobInfo(RunningJob rj) {
        if (ShimLoader.getHadoopShims().isLocalMode(job)) {
            console.printInfo("Job running in-process (local Hadoop)");
        } else {
            if (SessionState.get() != null) {
                SessionState.get().getHiveHistory().setTaskProperty(SessionState.get().getQueryId(), getId(),
                        Keys.TASK_HADOOP_ID, rj.getID().toString());
            }
            console.printInfo(getJobStartMsg(rj.getID()) + ", Tracking URL = " + rj.getTrackingURL());
            console.printInfo("Kill Command = " + HiveConf.getVar(job, HiveConf.ConfVars.HADOOPBIN) + " job  -kill "
                    + rj.getID());
        }
    }

    /**
     * This class contains the state of the running task Going forward, we will return this handle
     * from execute and Driver can split execute into start, monitorProgess and postProcess.
     */
    private static class ExecDriverTaskHandle extends TaskHandle {
        JobClient jc;
        RunningJob rj;
        HiveTxnManager txnMgr;

        JobClient getJobClient() {
            return jc;
        }

        RunningJob getRunningJob() {
            return rj;
        }

        HiveTxnManager getTxnManager() {
            return txnMgr;
        }

        public ExecDriverTaskHandle(JobClient jc, RunningJob rj, HiveTxnManager txnMgr) {
            this.jc = jc;
            this.rj = rj;
            this.txnMgr = txnMgr;
        }

        public void setRunningJob(RunningJob job) {
            rj = job;
        }

        @Override
        public Counters getCounters() throws IOException {
            return rj.getCounters();
        }
    }

    public void localJobDebugger(int exitVal, String taskId) {
        StringBuilder sb = new StringBuilder();
        sb.append("\n");
        sb.append("Task failed!\n");
        sb.append("Task ID:\n  " + taskId + "\n\n");
        sb.append("Logs:\n");
        console.printError(sb.toString());

        for (Appender a : Collections.list((Enumeration<Appender>) LogManager.getRootLogger().getAllAppenders())) {
            if (a instanceof FileAppender) {
                console.printError((new Path(((FileAppender) a).getFile())).toUri().getPath());
            }
        }
    }

    public int progressLocal(Process runningJob, String taskId) {
        int exitVal = -101;
        try {
            exitVal = runningJob.waitFor(); //TODO: poll periodically
        } catch (InterruptedException e) {
        }

        if (exitVal != 0) {
            console.printError("Execution failed with exit status: " + exitVal);
            console.printError("Obtaining error information");
            if (HiveConf.getBoolVar(job, HiveConf.ConfVars.SHOW_JOB_FAIL_DEBUG_INFO)) {
                // Since local jobs are run sequentially, all relevant information is already available
                // Therefore, no need to fetch job debug info asynchronously
                localJobDebugger(exitVal, taskId);
            }
        } else {
            console.printInfo("Execution completed successfully");
            console.printInfo("MapredLocal task succeeded");
        }
        return exitVal;
    }

    public int progress(RunningJob rj, JobClient jc, HiveTxnManager txnMgr) throws IOException {
        jobId = rj.getID();

        int returnVal = 0;

        // remove the pwd from conf file so that job tracker doesn't show this
        // logs
        String pwd = HiveConf.getVar(job, HiveConf.ConfVars.METASTOREPWD);
        if (pwd != null) {
            HiveConf.setVar(job, HiveConf.ConfVars.METASTOREPWD, "HIVE");
        }

        // replace it back
        if (pwd != null) {
            HiveConf.setVar(job, HiveConf.ConfVars.METASTOREPWD, pwd);
        }

        // add to list of running jobs to kill in case of abnormal shutdown

        runningJobs.add(rj);

        ExecDriverTaskHandle th = new ExecDriverTaskHandle(jc, rj, txnMgr);
        jobInfo(rj);
        MapRedStats mapRedStats = progress(th);

        this.task.taskHandle = th;
        // Not always there is a SessionState. Sometimes ExeDriver is directly invoked
        // for special modes. In that case, SessionState.get() is empty.
        if (SessionState.get() != null) {
            SessionState.get().getMapRedStats().put(getId(), mapRedStats);

            // Computes the skew for all the MapReduce irrespective
            // of Success or Failure
            if (this.task.getQueryPlan() != null) {
                computeReducerTimeStatsPerJob(rj);
            }
        }

        boolean success = mapRedStats.isSuccess();

        String statusMesg = getJobEndMsg(rj.getID());
        if (!success) {
            statusMesg += " with errors";
            returnVal = 2;
            console.printError(statusMesg);
            if (HiveConf.getBoolVar(job, HiveConf.ConfVars.SHOW_JOB_FAIL_DEBUG_INFO)
                    || HiveConf.getBoolVar(job, HiveConf.ConfVars.JOB_DEBUG_CAPTURE_STACKTRACES)) {
                try {
                    JobDebugger jd;
                    if (SessionState.get() != null) {
                        jd = new JobDebugger(job, rj, console, SessionState.get().getStackTraces());
                    } else {
                        jd = new JobDebugger(job, rj, console);
                    }
                    Thread t = new Thread(jd);
                    t.start();
                    t.join(HiveConf.getIntVar(job, HiveConf.ConfVars.JOB_DEBUG_TIMEOUT));
                    int ec = jd.getErrorCode();
                    if (ec > 0) {
                        returnVal = ec;
                    }
                } catch (InterruptedException e) {
                    console.printError("Timed out trying to grab more detailed job failure"
                            + " information, please check jobtracker for more info");
                }
            }
        } else {
            console.printInfo(statusMesg);
        }

        return returnVal;
    }

    private void computeReducerTimeStatsPerJob(RunningJob rj) throws IOException {
        TaskCompletionEvent[] taskCompletions = rj.getTaskCompletionEvents(0);
        List<Integer> reducersRunTimes = new ArrayList<Integer>();

        for (TaskCompletionEvent taskCompletion : taskCompletions) {
            if (!taskCompletion.isMapTask()) {
                reducersRunTimes.add(new Integer(taskCompletion.getTaskRunTime()));
            }
        }
        // Compute the reducers run time statistics for the job
        ReducerTimeStatsPerJob reducerTimeStatsPerJob = new ReducerTimeStatsPerJob(reducersRunTimes);
        // Adding the reducers run time statistics for the job in the QueryPlan
        this.task.getQueryPlan().getReducerTimeStatsPerJobList().add(reducerTimeStatsPerJob);
        return;
    }

    private Map<String, Double> extractAllCounterValues(Counters counters) {
        Map<String, Double> exctractedCounters = new HashMap<String, Double>();
        for (Counters.Group cg : counters) {
            for (Counter c : cg) {
                exctractedCounters.put(cg.getName() + "::" + c.getName(), new Double(c.getCounter()));
            }
        }
        return exctractedCounters;
    }

    private List<ClientStatsPublisher> getClientStatPublishers() {
        List<ClientStatsPublisher> clientStatsPublishers = new ArrayList<ClientStatsPublisher>();
        String confString = HiveConf.getVar(job, HiveConf.ConfVars.CLIENTSTATSPUBLISHERS);
        confString = confString.trim();
        if (confString.equals("")) {
            return clientStatsPublishers;
        }

        String[] clientStatsPublisherClasses = confString.split(",");

        for (String clientStatsPublisherClass : clientStatsPublisherClasses) {
            try {
                clientStatsPublishers.add((ClientStatsPublisher) Class
                        .forName(clientStatsPublisherClass.trim(), true, Utilities.getSessionSpecifiedClassLoader())
                        .newInstance());
            } catch (Exception e) {
                LOG.warn(e.getClass().getName() + " occured when trying to create class: "
                        + clientStatsPublisherClass.trim() + " implementing ClientStatsPublisher interface");
                LOG.warn("The exception message is: " + e.getMessage());
                LOG.warn("Program will continue, but without this ClientStatsPublisher working");
            }
        }
        return clientStatsPublishers;
    }
}