Package org.apache.hadoop.mapred.gridmix

Source Code of org.apache.hadoop.mapred.gridmix.TestGridmixSubmission$TestMonitor

/**
* 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.mapred.gridmix;

import java.io.InputStream;
import java.io.IOException;
import java.text.DecimalFormat;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.HashMap;
import java.util.concurrent.BlockingQueue;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.LinkedBlockingQueue;
import java.util.zip.GZIPInputStream;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.ContentSummary;
import org.apache.hadoop.fs.FileStatus;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.fs.permission.FsPermission;
import org.apache.hadoop.mapred.Counters;
import org.apache.hadoop.mapred.JobClient;
import org.apache.hadoop.mapred.JobConf;
import org.apache.hadoop.mapred.JobID;
import org.apache.hadoop.mapred.TaskReport;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.MRJobConfig;
import org.apache.hadoop.mapreduce.TaskType;
import org.apache.hadoop.tools.rumen.JobStory;
import org.apache.hadoop.tools.rumen.JobStoryProducer;
import org.apache.hadoop.tools.rumen.TaskInfo;
import org.apache.hadoop.util.ToolRunner;

import org.junit.AfterClass;
import org.junit.BeforeClass;
import org.junit.Test;
import static org.junit.Assert.*;

import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import org.apache.commons.logging.impl.Log4JLogger;
import org.apache.log4j.Level;

import static org.apache.hadoop.mapreduce.TaskCounter.MAP_INPUT_RECORDS;
import static org.apache.hadoop.mapreduce.TaskCounter.MAP_OUTPUT_BYTES;
import static org.apache.hadoop.mapreduce.TaskCounter.MAP_OUTPUT_RECORDS;
import static org.apache.hadoop.mapreduce.TaskCounter.REDUCE_INPUT_RECORDS;
import static org.apache.hadoop.mapreduce.TaskCounter.REDUCE_OUTPUT_RECORDS;
import static org.apache.hadoop.mapreduce.TaskCounter.REDUCE_SHUFFLE_BYTES;
import static org.apache.hadoop.mapreduce.TaskCounter.SPLIT_RAW_BYTES;

public class TestGridmixSubmission {
  static GridmixJobSubmissionPolicy policy = GridmixJobSubmissionPolicy.REPLAY;
  public static final Log LOG = LogFactory.getLog(Gridmix.class);

  {
    ((Log4JLogger)LogFactory.getLog("org.apache.hadoop.mapred.gridmix")
        ).getLogger().setLevel(Level.DEBUG);
  }

  private static final int NJOBS = 3;
  private static final long GENDATA = 30; // in megabytes
  private static final int GENSLOP = 100 * 1024; // +/- 100k for logs

  @BeforeClass
  public static void init() throws IOException {
    GridmixTestUtils.initCluster();
  }

  @AfterClass
  public static void shutDown() throws IOException {
    GridmixTestUtils.shutdownCluster();
  }

  static class TestMonitor extends JobMonitor {

    static final long SLOPBYTES = 1024;
    private final int expected;
    private final BlockingQueue<Job> retiredJobs;

    public TestMonitor(int expected, Statistics stats) {
      super(stats);
      this.expected = expected;
      retiredJobs = new LinkedBlockingQueue<Job>();
    }

    public void verify(ArrayList<JobStory> submitted, Configuration clientConf)
    throws Exception {
      final ArrayList<Job> succeeded = new ArrayList<Job>();
      assertEquals("Bad job count", expected, retiredJobs.drainTo(succeeded));
      final HashMap<String,JobStory> sub = new HashMap<String,JobStory>();
     
      // define the input and output path for the run
      final Path in = new Path("foo").makeQualified(GridmixTestUtils.dfs);
      final Path out =
        new Path(in, clientConf.get(Gridmix.GRIDMIX_OUT_DIR, "gridmix"));
     
      for (JobStory spec : submitted) {
        sub.put(spec.getJobID().toString(), spec);
      }
      final JobClient client = new JobClient(
        GridmixTestUtils.mrCluster.createJobConf());
      for (Job job : succeeded) {
        final String jobName = job.getJobName();
        Configuration conf = job.getConfiguration();
        if (GenerateData.JOB_NAME.equals(jobName)) {
          verifyQueue(conf, jobName);
         
          final ContentSummary generated = GridmixTestUtils.dfs.getContentSummary(in);
          assertTrue("Mismatched data gen", // +/- 100k for logs
              (GENDATA << 20) < generated.getLength() + GENSLOP ||
              (GENDATA << 20) > generated.getLength() - GENSLOP);
          FileStatus[] outstat = GridmixTestUtils.dfs.listStatus(out);
          assertEquals("Mismatched job count", NJOBS, outstat.length);
          continue;
        } else if (GenerateDistCacheData.JOB_NAME.equals(jobName)) {
          verifyQueue(conf, jobName);
          continue;
        }
       
        if (!conf.getBoolean(
          GridmixJob.GRIDMIX_USE_QUEUE_IN_TRACE, true)) {
          assertEquals(" Improper queue for  " + jobName + " " ,
              conf.get(MRJobConfig.QUEUE_NAME), "q1" );
        } else {
          assertEquals(" Improper queue for  " + jobName + " ",
              conf.get(MRJobConfig.QUEUE_NAME),
              sub.get(conf.get(Gridmix.ORIGINAL_JOB_ID)).getQueueName());
        }

        final String originalJobId = conf.get(Gridmix.ORIGINAL_JOB_ID);
        final JobStory spec = sub.get(originalJobId);
        assertNotNull("No spec for " + jobName, spec);
        assertNotNull("No counters for " + jobName, job.getCounters());
        final String originalJobName = spec.getName();
        System.out.println("originalJobName=" + originalJobName
            + ";GridmixJobName=" + jobName + ";originalJobID=" + originalJobId);
        assertTrue("Original job name is wrong.", originalJobName.equals(
            conf.get(Gridmix.ORIGINAL_JOB_NAME)));

        // Gridmix job seqNum contains 6 digits
        int seqNumLength = 6;
        String jobSeqNum = new DecimalFormat("000000").format(
            conf.getInt(GridmixJob.GRIDMIX_JOB_SEQ, -1));
        // Original job name is of the format MOCKJOB<6 digit sequence number>
        // because MockJob jobNames are of this format.
        assertTrue(originalJobName.substring(
            originalJobName.length() - seqNumLength).equals(jobSeqNum));

        assertTrue("Gridmix job name is not in the expected format.",
            jobName.equals(
            GridmixJob.JOB_NAME_PREFIX + jobSeqNum));

        final FileStatus stat =
          GridmixTestUtils.dfs.getFileStatus(
            new Path(out, "" + Integer.valueOf(jobSeqNum)));
        assertEquals("Wrong owner for " + jobName, spec.getUser(),
                     stat.getOwner());

        final int nMaps = spec.getNumberMaps();
        final int nReds = spec.getNumberReduces();

        // TODO Blocked by MAPREDUCE-118
        if (true) return;
        // TODO
        System.out.println(jobName + ": " + nMaps + "/" + nReds);
        final TaskReport[] mReports =
          client.getMapTaskReports(JobID.downgrade(job.getJobID()));
        assertEquals("Mismatched map count", nMaps, mReports.length);
        check(TaskType.MAP, job, spec, mReports,
            0, 0, SLOPBYTES, nReds);

        final TaskReport[] rReports =
          client.getReduceTaskReports(JobID.downgrade(job.getJobID()));
        assertEquals("Mismatched reduce count", nReds, rReports.length);
        check(TaskType.REDUCE, job, spec, rReports,
            nMaps * SLOPBYTES, 2 * nMaps, 0, 0);
      }
    }

    // Verify if correct job queue is used
    private void verifyQueue(Configuration conf, String jobName) {
      if (!conf.getBoolean(
          GridmixJob.GRIDMIX_USE_QUEUE_IN_TRACE, true)) {
        assertEquals(" Improper queue for " + jobName,
            conf.get("mapred.job.queue.name"), "q1");
      } else {
        assertEquals(" Improper queue for " + jobName,
            conf.get("mapred.job.queue.name"), "default");
      }
    }

    public void check(final TaskType type, Job job, JobStory spec,
          final TaskReport[] runTasks,
          long extraInputBytes, int extraInputRecords,
          long extraOutputBytes, int extraOutputRecords) throws Exception {

      long[] runInputRecords = new long[runTasks.length];
      long[] runInputBytes = new long[runTasks.length];
      long[] runOutputRecords = new long[runTasks.length];
      long[] runOutputBytes = new long[runTasks.length];
      long[] specInputRecords = new long[runTasks.length];
      long[] specInputBytes = new long[runTasks.length];
      long[] specOutputRecords = new long[runTasks.length];
      long[] specOutputBytes = new long[runTasks.length];

      for (int i = 0; i < runTasks.length; ++i) {
        final TaskInfo specInfo;
        final Counters counters = runTasks[i].getCounters();
        switch (type) {
          case MAP:
             runInputBytes[i] = counters.findCounter("FileSystemCounters",
                 "HDFS_BYTES_READ").getValue() -
                 counters.findCounter(SPLIT_RAW_BYTES).getValue();
             runInputRecords[i] =
               (int)counters.findCounter(MAP_INPUT_RECORDS).getValue();
             runOutputBytes[i] =
               counters.findCounter(MAP_OUTPUT_BYTES).getValue();
             runOutputRecords[i] =
               (int)counters.findCounter(MAP_OUTPUT_RECORDS).getValue();

            specInfo = spec.getTaskInfo(TaskType.MAP, i);
            specInputRecords[i] = specInfo.getInputRecords();
            specInputBytes[i] = specInfo.getInputBytes();
            specOutputRecords[i] = specInfo.getOutputRecords();
            specOutputBytes[i] = specInfo.getOutputBytes();
            System.out.printf(type + " SPEC: %9d -> %9d :: %5d -> %5d\n",
                 specInputBytes[i], specOutputBytes[i],
                 specInputRecords[i], specOutputRecords[i]);
            System.out.printf(type + " RUN:  %9d -> %9d :: %5d -> %5d\n",
                 runInputBytes[i], runOutputBytes[i],
                 runInputRecords[i], runOutputRecords[i]);
            break;
          case REDUCE:
            runInputBytes[i] = 0;
            runInputRecords[i] =
              (int)counters.findCounter(REDUCE_INPUT_RECORDS).getValue();
            runOutputBytes[i] =
              counters.findCounter("FileSystemCounters",
                  "HDFS_BYTES_WRITTEN").getValue();
            runOutputRecords[i] =
              (int)counters.findCounter(REDUCE_OUTPUT_RECORDS).getValue();


            specInfo = spec.getTaskInfo(TaskType.REDUCE, i);
            // There is no reliable counter for reduce input bytes. The
            // variable-length encoding of intermediate records and other noise
            // make this quantity difficult to estimate. The shuffle and spec
            // input bytes are included in debug output for reference, but are
            // not checked
            specInputBytes[i] = 0;
            specInputRecords[i] = specInfo.getInputRecords();
            specOutputRecords[i] = specInfo.getOutputRecords();
            specOutputBytes[i] = specInfo.getOutputBytes();
            System.out.printf(type + " SPEC: (%9d) -> %9d :: %5d -> %5d\n",
                 specInfo.getInputBytes(), specOutputBytes[i],
                 specInputRecords[i], specOutputRecords[i]);
            System.out.printf(type + " RUN:  (%9d) -> %9d :: %5d -> %5d\n",
                 counters.findCounter(REDUCE_SHUFFLE_BYTES).getValue(),
                 runOutputBytes[i], runInputRecords[i], runOutputRecords[i]);
            break;
          default:
            specInfo = null;
            fail("Unexpected type: " + type);
        }
      }

      // Check input bytes
      Arrays.sort(specInputBytes);
      Arrays.sort(runInputBytes);
      for (int i = 0; i < runTasks.length; ++i) {
        assertTrue("Mismatched " + type + " input bytes " +
            specInputBytes[i] + "/" + runInputBytes[i],
            eqPlusMinus(runInputBytes[i], specInputBytes[i], extraInputBytes));
      }

      // Check input records
      Arrays.sort(specInputRecords);
      Arrays.sort(runInputRecords);
      for (int i = 0; i < runTasks.length; ++i) {
        assertTrue("Mismatched " + type + " input records " +
            specInputRecords[i] + "/" + runInputRecords[i],
            eqPlusMinus(runInputRecords[i], specInputRecords[i],
              extraInputRecords));
      }

      // Check output bytes
      Arrays.sort(specOutputBytes);
      Arrays.sort(runOutputBytes);
      for (int i = 0; i < runTasks.length; ++i) {
        assertTrue("Mismatched " + type + " output bytes " +
            specOutputBytes[i] + "/" + runOutputBytes[i],
            eqPlusMinus(runOutputBytes[i], specOutputBytes[i],
              extraOutputBytes));
      }

      // Check output records
      Arrays.sort(specOutputRecords);
      Arrays.sort(runOutputRecords);
      for (int i = 0; i < runTasks.length; ++i) {
        assertTrue("Mismatched " + type + " output records " +
            specOutputRecords[i] + "/" + runOutputRecords[i],
            eqPlusMinus(runOutputRecords[i], specOutputRecords[i],
              extraOutputRecords));
      }

    }

    private static boolean eqPlusMinus(long a, long b, long x) {
      final long diff = Math.abs(a - b);
      return diff <= x;
    }

    @Override
    protected void onSuccess(Job job) {
      retiredJobs.add(job);
    }
    @Override
    protected void onFailure(Job job) {
      fail("Job failure: " + job);
    }
  }

  static class DebugGridmix extends Gridmix {

    private JobFactory factory;
    private TestMonitor monitor;

    public void checkMonitor(Configuration conf) throws Exception {
      monitor.verify(((DebugJobFactory.Debuggable)factory).getSubmitted(),
                     conf);
    }

    @Override
    protected JobMonitor createJobMonitor(Statistics stats) {
      monitor = new TestMonitor(NJOBS + 1, stats);
      return monitor;
    }

    @Override
    protected JobFactory createJobFactory(JobSubmitter submitter,
        String traceIn, Path scratchDir, Configuration conf,
        CountDownLatch startFlag, UserResolver userResolver)
        throws IOException {
      factory = DebugJobFactory.getFactory(
        submitter, scratchDir, NJOBS, conf, startFlag, userResolver);
      return factory;
    }
  }

  /**
   * Verifies that the given {@code JobStory} corresponds to the checked-in
   * WordCount {@code JobStory}. The verification is effected via JUnit
   * assertions.
   *
   * @param js the candidate JobStory.
   */
  private void verifyWordCountJobStory(JobStory js) {
    assertNotNull("Null JobStory", js);
    String expectedJobStory = "WordCount:johndoe:default:1285322645148:3:1";
    String actualJobStory = js.getName() + ":" + js.getUser() + ":"
      + js.getQueueName() + ":" + js.getSubmissionTime() + ":"
      + js.getNumberMaps() + ":" + js.getNumberReduces();
    assertEquals("Unexpected JobStory", expectedJobStory, actualJobStory);
  }

  /**
   * Expands a file compressed using {@code gzip}.
   *
   * @param fs the {@code FileSystem} corresponding to the given
   * file.
   *
   * @param in the path to the compressed file.
   *
   * @param out the path to the uncompressed output.
   *
   * @throws Exception if there was an error during the operation.
   */
  private void expandGzippedTrace(FileSystem fs, Path in, Path out)
    throws Exception {
    byte[] buff = new byte[4096];
    GZIPInputStream gis = new GZIPInputStream(fs.open(in));
    FSDataOutputStream fsdos = fs.create(out);
    int numRead;
    while ((numRead = gis.read(buff, 0, buff.length)) != -1) {
      fsdos.write(buff, 0, numRead);
    }
    gis.close();
    fsdos.close();
  }

  /**
   * Tests the reading of traces in GridMix3. These traces are generated
   * by Rumen and are in the JSON format. The traces can optionally be
   * compressed and uncompressed traces can also be passed to GridMix3 via
   * its standard input stream. The testing is effected via JUnit assertions.
   *
   * @throws Exception if there was an error.
   */
  @Test
  public void testTraceReader() throws Exception {
    Configuration conf = new Configuration();
    FileSystem lfs = FileSystem.getLocal(conf);
    Path rootInputDir = new Path(System.getProperty("src.test.data"));
    rootInputDir
      = rootInputDir.makeQualified(lfs.getUri(), lfs.getWorkingDirectory());
    Path rootTempDir
      = new Path(System.getProperty("test.build.data",
        System.getProperty("java.io.tmpdir")), "testTraceReader");
    rootTempDir
      = rootTempDir.makeQualified(lfs.getUri(), lfs.getWorkingDirectory());
    Path inputFile = new Path(rootInputDir, "wordcount.json.gz");
    Path tempFile = new Path(rootTempDir, "gridmix3-wc.json");

    InputStream origStdIn = System.in;
    InputStream tmpIs = null;
    try {
      DebugGridmix dgm = new DebugGridmix();
      JobStoryProducer jsp
        = dgm.createJobStoryProducer(inputFile.toString(), conf);

      System.out.println("Verifying JobStory from compressed trace...");
      verifyWordCountJobStory(jsp.getNextJob());

      expandGzippedTrace(lfs, inputFile, tempFile);
      jsp = dgm.createJobStoryProducer(tempFile.toString(), conf);
      System.out.println("Verifying JobStory from uncompressed trace...");
      verifyWordCountJobStory(jsp.getNextJob());

      tmpIs = lfs.open(tempFile);
      System.setIn(tmpIs);
      System.out.println("Verifying JobStory from trace in standard input...");
      jsp = dgm.createJobStoryProducer("-", conf);
      verifyWordCountJobStory(jsp.getNextJob());
    } finally {
      System.setIn(origStdIn);
      if (tmpIs != null) {
  tmpIs.close();
      }
      lfs.delete(rootTempDir, true);
    }
  }

  @Test
  public void testReplaySubmit() throws Exception {
    policy = GridmixJobSubmissionPolicy.REPLAY;
    System.out.println(" Replay started at " + System.currentTimeMillis());
    doSubmission(false, false);
    System.out.println(" Replay ended at " + System.currentTimeMillis());

    System.out.println(" Replay started with default output path at time "
        + System.currentTimeMillis());
    doSubmission(false, true);
    System.out.println(" Replay ended with default output path at time "
        + System.currentTimeMillis());
  }

  @Test
  public void testStressSubmit() throws Exception {
    policy = GridmixJobSubmissionPolicy.STRESS;
    System.out.println(" Stress started at " + System.currentTimeMillis());
    doSubmission(false, false);
    System.out.println(" Stress ended at " + System.currentTimeMillis());
  }

  @Test
  public void testStressSubmitWithDefaultQueue() throws Exception {
    policy = GridmixJobSubmissionPolicy.STRESS;
    System.out.println(" Stress with default q started at "
                       + System.currentTimeMillis());
    doSubmission(true, false);
    System.out.println(" Stress with default q ended at "
                       + System.currentTimeMillis());
  }

  @Test
  public void testSerialSubmit() throws Exception {
    policy = GridmixJobSubmissionPolicy.SERIAL;
    System.out.println("Serial started at " + System.currentTimeMillis());
    doSubmission(false, false);
    System.out.println("Serial ended at " + System.currentTimeMillis());
  }

  private void doSubmission(boolean useDefaultQueue,
      boolean defaultOutputPath) throws Exception {
    final Path in = new Path("foo").makeQualified(GridmixTestUtils.dfs);
    final Path out = GridmixTestUtils.DEST.makeQualified(GridmixTestUtils.dfs);
    final Path root = new Path("/user");
    Configuration conf = null;

    try{
      ArrayList<String> argsList = new ArrayList<String>();

      argsList.add("-D" + FilePool.GRIDMIX_MIN_FILE + "=0");
      argsList.add("-D" + Gridmix.GRIDMIX_USR_RSV + "="
          + EchoUserResolver.class.getName());

      // Set the config property gridmix.output.directory only if
      // defaultOutputPath is false. If defaultOutputPath is true, then
      // let us allow gridmix to use the path foo/gridmix/ as output dir.
      if (!defaultOutputPath) {
        argsList.add("-D" + Gridmix.GRIDMIX_OUT_DIR + "=" + out);
      }
      argsList.add("-generate");
      argsList.add(String.valueOf(GENDATA) + "m");
      argsList.add(in.toString());
      argsList.add("-"); // ignored by DebugGridmix

      String[] argv = argsList.toArray(new String[argsList.size()]);

      DebugGridmix client = new DebugGridmix();
      conf = new Configuration();
      conf.setEnum(GridmixJobSubmissionPolicy.JOB_SUBMISSION_POLICY,policy);
      conf.set("mapreduce.job.hdfs-servers", "");
      if (useDefaultQueue) {
        conf.setBoolean(GridmixJob.GRIDMIX_USE_QUEUE_IN_TRACE, false);
        conf.set(GridmixJob.GRIDMIX_DEFAULT_QUEUE, "q1");
      } else {
        conf.setBoolean(GridmixJob.GRIDMIX_USE_QUEUE_IN_TRACE, true);
      }
      conf = GridmixTestUtils.mrCluster.createJobConf(new JobConf(conf));
      // allow synthetic users to create home directories
      GridmixTestUtils.dfs.mkdirs(root, new FsPermission((short)0777));
      GridmixTestUtils.dfs.setPermission(root, new FsPermission((short)0777));
      int res = ToolRunner.run(conf, client, argv);
      assertEquals("Client exited with nonzero status", 0, res);
      client.checkMonitor(conf);
    } catch (Exception e) {
      e.printStackTrace();
      // fail the test if there is an exception
      throw new RuntimeException(e);
    } finally {
      in.getFileSystem(conf).delete(in, true);
      out.getFileSystem(conf).delete(out, true);
      root.getFileSystem(conf).delete(root,true);
    }
  }

}
TOP

Related Classes of org.apache.hadoop.mapred.gridmix.TestGridmixSubmission$TestMonitor

TOP
Copyright © 2018 www.massapi.com. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.