Java JobContext类代码示例

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@Test
public void testNotificationOnLastRetryNormalShutdown() throws Exception {
  HttpServer2 server = startHttpServer();
  // Act like it is the second attempt. Default max attempts is 2
  MRApp app = spy(new MRAppWithCustomContainerAllocator(
      2, 2, true, this.getClass().getName(), true, 2, true));
  doNothing().when(app).sysexit();
  JobConf conf = new JobConf();
  conf.set(JobContext.MR_JOB_END_NOTIFICATION_URL,
      JobEndServlet.baseUrl + "jobend?jobid=$jobId&status=$jobStatus");
  JobImpl job = (JobImpl)app.submit(conf);
  app.waitForInternalState(job, JobStateInternal.SUCCEEDED);
  // Unregistration succeeds: successfullyUnregistered is set
  app.shutDownJob();
  Assert.assertTrue(app.isLastAMRetry());
  Assert.assertEquals(1, JobEndServlet.calledTimes);
  Assert.assertEquals("jobid=" + job.getID() + "&status=SUCCEEDED",
      JobEndServlet.requestUri.getQuery());
  Assert.assertEquals(JobState.SUCCEEDED.toString(),
    JobEndServlet.foundJobState);
  server.stop();
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@Test
public void testAbsentNotificationOnNotLastRetryUnregistrationFailure()
    throws Exception {
  HttpServer2 server = startHttpServer();
  MRApp app = spy(new MRAppWithCustomContainerAllocator(2, 2, false,
      this.getClass().getName(), true, 1, false));
  doNothing().when(app).sysexit();
  JobConf conf = new JobConf();
  conf.set(JobContext.MR_JOB_END_NOTIFICATION_URL,
      JobEndServlet.baseUrl + "jobend?jobid=$jobId&status=$jobStatus");
  JobImpl job = (JobImpl)app.submit(conf);
  app.waitForState(job, JobState.RUNNING);
  app.getContext().getEventHandler()
    .handle(new JobEvent(app.getJobId(), JobEventType.JOB_AM_REBOOT));
  app.waitForInternalState(job, JobStateInternal.REBOOT);
  // Now shutdown.
  // Unregistration fails: isLastAMRetry is recalculated, this is not
  app.shutDownJob();
  // Not the last AM attempt. So user should that the job is still running.
  app.waitForState(job, JobState.RUNNING);
  Assert.assertFalse(app.isLastAMRetry());
  Assert.assertEquals(0, JobEndServlet.calledTimes);
  Assert.assertNull(JobEndServlet.requestUri);
  Assert.assertNull(JobEndServlet.foundJobState);
  server.stop();
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
public void configure(String keySpec, int expect) throws Exception {
  Path testdir = new Path(TEST_DIR.getAbsolutePath());
  Path inDir = new Path(testdir, "in");
  Path outDir = new Path(testdir, "out");
  FileSystem fs = getFileSystem();
  fs.delete(testdir, true);
  conf.setInputFormat(TextInputFormat.class);
  FileInputFormat.setInputPaths(conf, inDir);
  FileOutputFormat.setOutputPath(conf, outDir);
  conf.setOutputKeyClass(Text.class);
  conf.setOutputValueClass(LongWritable.class);

  conf.setNumMapTasks(1);
  conf.setNumReduceTasks(1);

  conf.setOutputFormat(TextOutputFormat.class);
  conf.setOutputKeyComparatorClass(KeyFieldBasedComparator.class);
  conf.setKeyFieldComparatorOptions(keySpec);
  conf.setKeyFieldPartitionerOptions("-k1.1,1.1");
  conf.set(JobContext.MAP_OUTPUT_KEY_FIELD_SEPERATOR, " ");
  conf.setMapperClass(InverseMapper.class);
  conf.setReducerClass(IdentityReducer.class);
  if (!fs.mkdirs(testdir)) {
    throw new IOException("Mkdirs failed to create " + testdir.toString());
  }
  if (!fs.mkdirs(inDir)) {
    throw new IOException("Mkdirs failed to create " + inDir.toString());
  }
  // set up input data in 2 files 
  Path inFile = new Path(inDir, "part0");
  FileOutputStream fos = new FileOutputStream(inFile.toString());
  fos.write((line1 + "\n").getBytes());
  fos.write((line2 + "\n").getBytes());
  fos.close();
  JobClient jc = new JobClient(conf);
  RunningJob r_job = jc.submitJob(conf);
  while (!r_job.isComplete()) {
    Thread.sleep(1000);
  }
  
  if (!r_job.isSuccessful()) {
    fail("Oops! The job broke due to an unexpected error");
  }
  Path[] outputFiles = FileUtil.stat2Paths(
      getFileSystem().listStatus(outDir,
      new Utils.OutputFileUtils.OutputFilesFilter()));
  if (outputFiles.length > 0) {
    InputStream is = getFileSystem().open(outputFiles[0]);
    BufferedReader reader = new BufferedReader(new InputStreamReader(is));
    String line = reader.readLine();
    //make sure we get what we expect as the first line, and also
    //that we have two lines
    if (expect == 1) {
      assertTrue(line.startsWith(line1));
    } else if (expect == 2) {
      assertTrue(line.startsWith(line2));
    }
    line = reader.readLine();
    if (expect == 1) {
      assertTrue(line.startsWith(line2));
    } else if (expect == 2) {
      assertTrue(line.startsWith(line1));
    }
    reader.close();
  }
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@SuppressWarnings({"rawtypes", "unchecked"})
@Test (timeout=10000)
public void testLoadMapper() throws Exception {

  Configuration conf = new Configuration();
  conf.setInt(JobContext.NUM_REDUCES, 2);

  CompressionEmulationUtil.setCompressionEmulationEnabled(conf, true);
  conf.setBoolean(MRJobConfig.MAP_OUTPUT_COMPRESS, true);

  TaskAttemptID taskId = new TaskAttemptID();
  RecordReader<NullWritable, GridmixRecord> reader = new FakeRecordReader();

  LoadRecordGkGrWriter writer = new LoadRecordGkGrWriter();

  OutputCommitter committer = new CustomOutputCommitter();
  StatusReporter reporter = new TaskAttemptContextImpl.DummyReporter();
  LoadSplit split = getLoadSplit();

  MapContext<NullWritable, GridmixRecord, GridmixKey, GridmixRecord> mapContext = new MapContextImpl<NullWritable, GridmixRecord, GridmixKey, GridmixRecord>(
          conf, taskId, reader, writer, committer, reporter, split);
  // context
  Context ctx = new WrappedMapper<NullWritable, GridmixRecord, GridmixKey, GridmixRecord>()
          .getMapContext(mapContext);

  reader.initialize(split, ctx);
  ctx.getConfiguration().setBoolean(MRJobConfig.MAP_OUTPUT_COMPRESS, true);
  CompressionEmulationUtil.setCompressionEmulationEnabled(
          ctx.getConfiguration(), true);

  LoadJob.LoadMapper mapper = new LoadJob.LoadMapper();
  // setup, map, clean
  mapper.run(ctx);

  Map<GridmixKey, GridmixRecord> data = writer.getData();
  // check result
  assertEquals(2, data.size());

}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@Test
public void testOpen() throws Exception {

	OutputFormat<String, Long> dummyOutputFormat = mock(DummyOutputFormat.class);
	DummyOutputCommitter outputCommitter = mock(DummyOutputCommitter.class);
	JobConf jobConf = Mockito.spy(new JobConf());
	when(jobConf.getOutputCommitter()).thenReturn(outputCommitter);

	HadoopOutputFormat<String, Long> outputFormat = new HadoopOutputFormat<>(dummyOutputFormat, jobConf);

	outputFormat.open(1, 1);

	verify(jobConf, times(2)).getOutputCommitter();
	verify(outputCommitter, times(1)).setupJob(any(JobContext.class));
	verify(dummyOutputFormat, times(1)).getRecordWriter(any(FileSystem.class), any(JobConf.class), anyString(), any(Progressable.class));
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@Test
public void testOpen() throws Exception {

	OutputFormat<String, Long> dummyOutputFormat = mock(DummyOutputFormat.class);
	DummyOutputCommitter outputCommitter = mock(DummyOutputCommitter.class);
	JobConf jobConf = spy(new JobConf());
	when(jobConf.getOutputCommitter()).thenReturn(outputCommitter);

	HadoopOutputFormat<String, Long> outputFormat = new HadoopOutputFormat<>(dummyOutputFormat, jobConf);

	outputFormat.open(1, 1);

	verify(jobConf, times(2)).getOutputCommitter();
	verify(outputCommitter, times(1)).setupJob(any(JobContext.class));
	verify(dummyOutputFormat, times(1)).getRecordWriter(any(FileSystem.class), any(JobConf.class), anyString(), any(Progressable.class));
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
/** {@inheritDoc} */
@Override public void run(HadoopTaskContext taskCtx) throws IgniteCheckedException {
    HadoopV2TaskContext ctx = (HadoopV2TaskContext)taskCtx;

    JobContext jobCtx = ctx.jobContext();

    try {
        OutputCommitter committer = jobCtx.getJobConf().getOutputCommitter();

        if (abort)
            committer.abortJob(jobCtx, JobStatus.State.FAILED);
        else
            committer.commitJob(jobCtx);
    }
    catch (IOException e) {
        throw new IgniteCheckedException(e);
    }
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@Test
public void testNotificationOnLastRetryNormalShutdown() throws Exception {
  HttpServer server = startHttpServer();
  // Act like it is the second attempt. Default max attempts is 2
  MRApp app = spy(new MRAppWithCustomContainerAllocator(
      2, 2, true, this.getClass().getName(), true, 2, true));
  doNothing().when(app).sysexit();
  Configuration conf = new Configuration();
  conf.set(JobContext.MR_JOB_END_NOTIFICATION_URL,
      JobEndServlet.baseUrl + "jobend?jobid=$jobId&status=$jobStatus");
  JobImpl job = (JobImpl)app.submit(conf);
  app.waitForInternalState(job, JobStateInternal.SUCCEEDED);
  // Unregistration succeeds: successfullyUnregistered is set
  app.shutDownJob();
  Assert.assertEquals(true, app.isLastAMRetry());
  Assert.assertEquals(1, JobEndServlet.calledTimes);
  Assert.assertEquals("jobid=" + job.getID() + "&status=SUCCEEDED",
      JobEndServlet.requestUri.getQuery());
  Assert.assertEquals(JobState.SUCCEEDED.toString(),
    JobEndServlet.foundJobState);
  server.stop();
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@Test
public void testAbsentNotificationOnNotLastRetryUnregistrationFailure()
    throws Exception {
  HttpServer server = startHttpServer();
  MRApp app = spy(new MRAppWithCustomContainerAllocator(2, 2, false,
      this.getClass().getName(), true, 1, false));
  doNothing().when(app).sysexit();
  Configuration conf = new Configuration();
  conf.set(JobContext.MR_JOB_END_NOTIFICATION_URL,
      JobEndServlet.baseUrl + "jobend?jobid=$jobId&status=$jobStatus");
  JobImpl job = (JobImpl)app.submit(conf);
  app.waitForState(job, JobState.RUNNING);
  app.getContext().getEventHandler()
    .handle(new JobEvent(app.getJobId(), JobEventType.JOB_AM_REBOOT));
  app.waitForInternalState(job, JobStateInternal.REBOOT);
  // Now shutdown.
  // Unregistration fails: isLastAMRetry is recalculated, this is not
  app.shutDownJob();
  // Not the last AM attempt. So user should that the job is still running.
  app.waitForState(job, JobState.RUNNING);
  Assert.assertEquals(false, app.isLastAMRetry());
  Assert.assertEquals(0, JobEndServlet.calledTimes);
  Assert.assertEquals(null, JobEndServlet.requestUri);
  Assert.assertEquals(null, JobEndServlet.foundJobState);
  server.stop();
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@Test
public void testNotificationOnLastRetryUnregistrationFailure()
    throws Exception {
  HttpServer server = startHttpServer();
  MRApp app = spy(new MRAppWithCustomContainerAllocator(2, 2, false,
      this.getClass().getName(), true, 2, false));
  doNothing().when(app).sysexit();
  Configuration conf = new Configuration();
  conf.set(JobContext.MR_JOB_END_NOTIFICATION_URL,
      JobEndServlet.baseUrl + "jobend?jobid=$jobId&status=$jobStatus");
  JobImpl job = (JobImpl)app.submit(conf);
  app.waitForState(job, JobState.RUNNING);
  app.getContext().getEventHandler()
    .handle(new JobEvent(app.getJobId(), JobEventType.JOB_AM_REBOOT));
  app.waitForInternalState(job, JobStateInternal.REBOOT);
  // Now shutdown. User should see FAILED state.
  // Unregistration fails: isLastAMRetry is recalculated, this is
  app.shutDownJob();
  Assert.assertEquals(true, app.isLastAMRetry());
  Assert.assertEquals(1, JobEndServlet.calledTimes);
  Assert.assertEquals("jobid=" + job.getID() + "&status=FAILED",
      JobEndServlet.requestUri.getQuery());
  Assert.assertEquals(JobState.FAILED.toString(),
    JobEndServlet.foundJobState);
  server.stop();
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@Test
public void testNotificationOnLastRetryUnregistrationFailure()
    throws Exception {
  HttpServer2 server = startHttpServer();
  MRApp app = spy(new MRAppWithCustomContainerAllocator(2, 2, false,
      this.getClass().getName(), true, 2, false));
  doNothing().when(app).sysexit();
  JobConf conf = new JobConf();
  conf.set(JobContext.MR_JOB_END_NOTIFICATION_URL,
      JobEndServlet.baseUrl + "jobend?jobid=$jobId&status=$jobStatus");
  JobImpl job = (JobImpl)app.submit(conf);
  app.waitForState(job, JobState.RUNNING);
  app.getContext().getEventHandler()
    .handle(new JobEvent(app.getJobId(), JobEventType.JOB_AM_REBOOT));
  app.waitForInternalState(job, JobStateInternal.REBOOT);
  // Now shutdown. User should see FAILED state.
  // Unregistration fails: isLastAMRetry is recalculated, this is
  app.shutDownJob();
  Assert.assertTrue(app.isLastAMRetry());
  Assert.assertEquals(1, JobEndServlet.calledTimes);
  Assert.assertEquals("jobid=" + job.getID() + "&status=FAILED",
      JobEndServlet.requestUri.getQuery());
  Assert.assertEquals(JobState.FAILED.toString(),
    JobEndServlet.foundJobState);
  server.stop();
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
public static JobContext instantiateJobContext(JobConf jobConf, JobID jobId) throws Exception {
	try {
		// for Hadoop 1.xx
		Class<?> clazz = null;
		if(!TaskAttemptContext.class.isInterface()) { 
			clazz = Class.forName("org.apache.hadoop.mapred.JobContext", true, Thread.currentThread().getContextClassLoader());
		}
		// for Hadoop 2.xx
		else {
			clazz = Class.forName("org.apache.hadoop.mapred.JobContextImpl", true, Thread.currentThread().getContextClassLoader());
		}
		Constructor<?> constructor = clazz.getDeclaredConstructor(JobConf.class, org.apache.hadoop.mapreduce.JobID.class);
		// for Hadoop 1.xx
		constructor.setAccessible(true);
		JobContext context = (JobContext) constructor.newInstance(jobConf, jobId);
		
		return context;
	} catch(Exception e) {
		throw new Exception("Could not create instance of JobContext.", e);
	}
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
public void localizeConfiguration(JobConf jobConf)
    throws IOException, InterruptedException {
  jobConf.set(JobContext.TASK_ID, taskAttemptId.getTaskID().toString());
  jobConf.set(JobContext.TASK_ATTEMPT_ID, taskAttemptId.toString());
  jobConf.setInt(JobContext.TASK_PARTITION,
      taskAttemptId.getTaskID().getId());
  jobConf.set(JobContext.ID, taskAttemptId.getJobID().toString());
  
  jobConf.setBoolean(MRJobConfig.TASK_ISMAP, isMap);
  
  Path outputPath = FileOutputFormat.getOutputPath(jobConf);
  if (outputPath != null) {
    if ((committer instanceof FileOutputCommitter)) {
      FileOutputFormat.setWorkOutputPath(jobConf, 
        ((FileOutputCommitter)committer).getTaskAttemptPath(taskAttemptContext));
    } else {
      FileOutputFormat.setWorkOutputPath(jobConf, outputPath);
    }
  }
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@Test
public void testNotificationOnLastRetryUnregistrationFailure()
    throws Exception {
  HttpServer2 server = startHttpServer();
  MRApp app = spy(new MRAppWithCustomContainerAllocator(2, 2, false,
      this.getClass().getName(), true, 2, false));
  // Currently, we will have isLastRetry always equals to false at beginning
  // of MRAppMaster, except staging area exists or commit already started at 
  // the beginning.
  // Now manually set isLastRetry to true and this should reset to false when
  // unregister failed.
  app.isLastAMRetry = true;
  doNothing().when(app).sysexit();
  JobConf conf = new JobConf();
  conf.set(JobContext.MR_JOB_END_NOTIFICATION_URL,
      JobEndServlet.baseUrl + "jobend?jobid=$jobId&status=$jobStatus");
  JobImpl job = (JobImpl)app.submit(conf);
  app.waitForState(job, JobState.RUNNING);
  app.getContext().getEventHandler()
    .handle(new JobEvent(app.getJobId(), JobEventType.JOB_AM_REBOOT));
  app.waitForInternalState(job, JobStateInternal.REBOOT);
  // Now shutdown. User should see FAILED state.
  // Unregistration fails: isLastAMRetry is recalculated, this is
  ///reboot will stop service internally, we don't need to shutdown twice
  app.waitForServiceToStop(10000);
  Assert.assertFalse(app.isLastAMRetry());
  // Since it's not last retry, JobEndServlet didn't called
  Assert.assertEquals(0, JobEndServlet.calledTimes);
  Assert.assertNull(JobEndServlet.requestUri);
  Assert.assertNull(JobEndServlet.foundJobState);
  server.stop();
}
 

import org.apache.hadoop.mapred.JobContext; //导入依赖的package包/类
@SuppressWarnings({"unchecked", "rawtypes"})
@Test (timeout=30000)
public void testSleepMapper() throws Exception {
  SleepJob.SleepMapper test = new SleepJob.SleepMapper();

  Configuration conf = new Configuration();
  conf.setInt(JobContext.NUM_REDUCES, 2);

  CompressionEmulationUtil.setCompressionEmulationEnabled(conf, true);
  conf.setBoolean(MRJobConfig.MAP_OUTPUT_COMPRESS, true);
  TaskAttemptID taskId = new TaskAttemptID();
  FakeRecordLLReader reader = new FakeRecordLLReader();
  LoadRecordGkNullWriter writer = new LoadRecordGkNullWriter();
  OutputCommitter committer = new CustomOutputCommitter();
  StatusReporter reporter = new TaskAttemptContextImpl.DummyReporter();
  SleepSplit split = getSleepSplit();
  MapContext<LongWritable, LongWritable, GridmixKey, NullWritable> mapcontext = new MapContextImpl<LongWritable, LongWritable, GridmixKey, NullWritable>(
          conf, taskId, reader, writer, committer, reporter, split);
  Context context = new WrappedMapper<LongWritable, LongWritable, GridmixKey, NullWritable>()
          .getMapContext(mapcontext);

  long start = System.currentTimeMillis();
  LOG.info("start:" + start);
  LongWritable key = new LongWritable(start + 2000);
  LongWritable value = new LongWritable(start + 2000);
  // should slip 2 sec
  test.map(key, value, context);
  LOG.info("finish:" + System.currentTimeMillis());
  assertTrue(System.currentTimeMillis() >= (start + 2000));

  test.cleanup(context);
  assertEquals(1, writer.getData().size());
}