Search in sources :

Example 1 with TaskSplitMetaInfo

use of org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo in project hadoop by apache.

the class TestJobImpl method testMetaInfoSizeOverMax.

@Test
public void testMetaInfoSizeOverMax() throws Exception {
    Configuration conf = new Configuration();
    JobID jobID = JobID.forName("job_1234567890000_0001");
    JobId jobId = TypeConverter.toYarn(jobID);
    MRAppMetrics mrAppMetrics = MRAppMetrics.create();
    JobImpl job = new JobImpl(jobId, ApplicationAttemptId.newInstance(ApplicationId.newInstance(0, 0), 0), conf, mock(EventHandler.class), null, new JobTokenSecretManager(), new Credentials(), null, null, mrAppMetrics, null, true, null, 0, null, null, null, null);
    InitTransition initTransition = new InitTransition() {

        @Override
        protected TaskSplitMetaInfo[] createSplits(JobImpl job, JobId jobId) {
            throw new YarnRuntimeException(EXCEPTIONMSG);
        }
    };
    JobEvent mockJobEvent = mock(JobEvent.class);
    JobStateInternal jobSI = initTransition.transition(job, mockJobEvent);
    Assert.assertTrue("When init fails, return value from InitTransition.transition should equal NEW.", jobSI.equals(JobStateInternal.NEW));
    Assert.assertTrue("Job diagnostics should contain YarnRuntimeException", job.getDiagnostics().toString().contains("YarnRuntimeException"));
    Assert.assertTrue("Job diagnostics should contain " + EXCEPTIONMSG, job.getDiagnostics().toString().contains(EXCEPTIONMSG));
}
Also used : Configuration(org.apache.hadoop.conf.Configuration) JobStateInternal(org.apache.hadoop.mapreduce.v2.app.job.JobStateInternal) CommitterEventHandler(org.apache.hadoop.mapreduce.v2.app.commit.CommitterEventHandler) EventHandler(org.apache.hadoop.yarn.event.EventHandler) YarnRuntimeException(org.apache.hadoop.yarn.exceptions.YarnRuntimeException) InitTransition(org.apache.hadoop.mapreduce.v2.app.job.impl.JobImpl.InitTransition) JobEvent(org.apache.hadoop.mapreduce.v2.app.job.event.JobEvent) JobTokenSecretManager(org.apache.hadoop.mapreduce.security.token.JobTokenSecretManager) TaskSplitMetaInfo(org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo) MRAppMetrics(org.apache.hadoop.mapreduce.v2.app.metrics.MRAppMetrics) JobID(org.apache.hadoop.mapreduce.JobID) JobId(org.apache.hadoop.mapreduce.v2.api.records.JobId) Credentials(org.apache.hadoop.security.Credentials) Test(org.junit.Test)

Example 2 with TaskSplitMetaInfo

use of org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo in project hadoop by apache.

the class TestTaskAttempt method testAppDiognosticEventOnNewTask.

@Test
public void testAppDiognosticEventOnNewTask() throws Exception {
    ApplicationId appId = ApplicationId.newInstance(1, 2);
    ApplicationAttemptId appAttemptId = ApplicationAttemptId.newInstance(appId, 0);
    JobId jobId = MRBuilderUtils.newJobId(appId, 1);
    TaskId taskId = MRBuilderUtils.newTaskId(jobId, 1, TaskType.MAP);
    TaskAttemptId attemptId = MRBuilderUtils.newTaskAttemptId(taskId, 0);
    Path jobFile = mock(Path.class);
    MockEventHandler eventHandler = new MockEventHandler();
    TaskAttemptListener taListener = mock(TaskAttemptListener.class);
    when(taListener.getAddress()).thenReturn(new InetSocketAddress("localhost", 0));
    JobConf jobConf = new JobConf();
    jobConf.setClass("fs.file.impl", StubbedFS.class, FileSystem.class);
    jobConf.setBoolean("fs.file.impl.disable.cache", true);
    jobConf.set(JobConf.MAPRED_MAP_TASK_ENV, "");
    jobConf.set(MRJobConfig.APPLICATION_ATTEMPT_ID, "10");
    TaskSplitMetaInfo splits = mock(TaskSplitMetaInfo.class);
    when(splits.getLocations()).thenReturn(new String[] { "127.0.0.1" });
    AppContext appCtx = mock(AppContext.class);
    ClusterInfo clusterInfo = mock(ClusterInfo.class);
    Resource resource = mock(Resource.class);
    when(appCtx.getClusterInfo()).thenReturn(clusterInfo);
    when(resource.getMemorySize()).thenReturn(1024L);
    setupTaskAttemptFinishingMonitor(eventHandler, jobConf, appCtx);
    TaskAttemptImpl taImpl = new MapTaskAttemptImpl(taskId, 1, eventHandler, jobFile, 1, splits, jobConf, taListener, new Token(), new Credentials(), SystemClock.getInstance(), appCtx);
    NodeId nid = NodeId.newInstance("127.0.0.1", 0);
    ContainerId contId = ContainerId.newContainerId(appAttemptId, 3);
    Container container = mock(Container.class);
    when(container.getId()).thenReturn(contId);
    when(container.getNodeId()).thenReturn(nid);
    when(container.getNodeHttpAddress()).thenReturn("localhost:0");
    taImpl.handle(new TaskAttemptDiagnosticsUpdateEvent(attemptId, "Task got killed"));
    assertFalse("InternalError occurred trying to handle TA_DIAGNOSTICS_UPDATE on assigned task", eventHandler.internalError);
}
Also used : Path(org.apache.hadoop.fs.Path) TaskId(org.apache.hadoop.mapreduce.v2.api.records.TaskId) TaskAttemptId(org.apache.hadoop.mapreduce.v2.api.records.TaskAttemptId) TaskAttemptListener(org.apache.hadoop.mapreduce.v2.app.TaskAttemptListener) InetSocketAddress(java.net.InetSocketAddress) TaskAttemptDiagnosticsUpdateEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptDiagnosticsUpdateEvent) AppContext(org.apache.hadoop.mapreduce.v2.app.AppContext) Resource(org.apache.hadoop.yarn.api.records.Resource) MapTaskAttemptImpl(org.apache.hadoop.mapred.MapTaskAttemptImpl) Token(org.apache.hadoop.security.token.Token) ApplicationAttemptId(org.apache.hadoop.yarn.api.records.ApplicationAttemptId) ClusterInfo(org.apache.hadoop.mapreduce.v2.app.ClusterInfo) Container(org.apache.hadoop.yarn.api.records.Container) ContainerId(org.apache.hadoop.yarn.api.records.ContainerId) NodeId(org.apache.hadoop.yarn.api.records.NodeId) TaskSplitMetaInfo(org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo) ApplicationId(org.apache.hadoop.yarn.api.records.ApplicationId) MapTaskAttemptImpl(org.apache.hadoop.mapred.MapTaskAttemptImpl) JobConf(org.apache.hadoop.mapred.JobConf) JobId(org.apache.hadoop.mapreduce.v2.api.records.JobId) Credentials(org.apache.hadoop.security.Credentials) Test(org.junit.Test)

Example 3 with TaskSplitMetaInfo

use of org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo in project hadoop by apache.

the class TestTaskAttempt method testContainerKillWhileRunning.

@Test
public void testContainerKillWhileRunning() throws Exception {
    ApplicationId appId = ApplicationId.newInstance(1, 2);
    ApplicationAttemptId appAttemptId = ApplicationAttemptId.newInstance(appId, 0);
    JobId jobId = MRBuilderUtils.newJobId(appId, 1);
    TaskId taskId = MRBuilderUtils.newTaskId(jobId, 1, TaskType.MAP);
    TaskAttemptId attemptId = MRBuilderUtils.newTaskAttemptId(taskId, 0);
    Path jobFile = mock(Path.class);
    MockEventHandler eventHandler = new MockEventHandler();
    TaskAttemptListener taListener = mock(TaskAttemptListener.class);
    when(taListener.getAddress()).thenReturn(new InetSocketAddress("localhost", 0));
    JobConf jobConf = new JobConf();
    jobConf.setClass("fs.file.impl", StubbedFS.class, FileSystem.class);
    jobConf.setBoolean("fs.file.impl.disable.cache", true);
    jobConf.set(JobConf.MAPRED_MAP_TASK_ENV, "");
    jobConf.set(MRJobConfig.APPLICATION_ATTEMPT_ID, "10");
    TaskSplitMetaInfo splits = mock(TaskSplitMetaInfo.class);
    when(splits.getLocations()).thenReturn(new String[] { "127.0.0.1" });
    AppContext appCtx = mock(AppContext.class);
    ClusterInfo clusterInfo = mock(ClusterInfo.class);
    Resource resource = mock(Resource.class);
    when(appCtx.getClusterInfo()).thenReturn(clusterInfo);
    when(resource.getMemorySize()).thenReturn(1024L);
    TaskAttemptImpl taImpl = new MapTaskAttemptImpl(taskId, 1, eventHandler, jobFile, 1, splits, jobConf, taListener, new Token(), new Credentials(), SystemClock.getInstance(), appCtx);
    NodeId nid = NodeId.newInstance("127.0.0.2", 0);
    ContainerId contId = ContainerId.newContainerId(appAttemptId, 3);
    Container container = mock(Container.class);
    when(container.getId()).thenReturn(contId);
    when(container.getNodeId()).thenReturn(nid);
    when(container.getNodeHttpAddress()).thenReturn("localhost:0");
    taImpl.handle(new TaskAttemptEvent(attemptId, TaskAttemptEventType.TA_SCHEDULE));
    taImpl.handle(new TaskAttemptContainerAssignedEvent(attemptId, container, mock(Map.class)));
    taImpl.handle(new TaskAttemptContainerLaunchedEvent(attemptId, 0));
    assertEquals("Task attempt is not in running state", taImpl.getState(), TaskAttemptState.RUNNING);
    taImpl.handle(new TaskAttemptEvent(attemptId, TaskAttemptEventType.TA_KILL));
    assertFalse("InternalError occurred trying to handle TA_KILL", eventHandler.internalError);
    assertEquals("Task should be in KILL_CONTAINER_CLEANUP state", TaskAttemptStateInternal.KILL_CONTAINER_CLEANUP, taImpl.getInternalState());
}
Also used : Path(org.apache.hadoop.fs.Path) TaskId(org.apache.hadoop.mapreduce.v2.api.records.TaskId) TaskAttemptId(org.apache.hadoop.mapreduce.v2.api.records.TaskAttemptId) TaskAttemptListener(org.apache.hadoop.mapreduce.v2.app.TaskAttemptListener) InetSocketAddress(java.net.InetSocketAddress) AppContext(org.apache.hadoop.mapreduce.v2.app.AppContext) Resource(org.apache.hadoop.yarn.api.records.Resource) MapTaskAttemptImpl(org.apache.hadoop.mapred.MapTaskAttemptImpl) Token(org.apache.hadoop.security.token.Token) TaskAttemptEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptEvent) ApplicationAttemptId(org.apache.hadoop.yarn.api.records.ApplicationAttemptId) TaskAttemptContainerAssignedEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptContainerAssignedEvent) TaskAttemptContainerLaunchedEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptContainerLaunchedEvent) ClusterInfo(org.apache.hadoop.mapreduce.v2.app.ClusterInfo) Container(org.apache.hadoop.yarn.api.records.Container) ContainerId(org.apache.hadoop.yarn.api.records.ContainerId) NodeId(org.apache.hadoop.yarn.api.records.NodeId) TaskSplitMetaInfo(org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo) ApplicationId(org.apache.hadoop.yarn.api.records.ApplicationId) MapTaskAttemptImpl(org.apache.hadoop.mapred.MapTaskAttemptImpl) JobConf(org.apache.hadoop.mapred.JobConf) JobId(org.apache.hadoop.mapreduce.v2.api.records.JobId) Credentials(org.apache.hadoop.security.Credentials) Test(org.junit.Test)

Example 4 with TaskSplitMetaInfo

use of org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo in project hadoop by apache.

the class TestTaskAttempt method testHostResolveAttempt.

@Test
public void testHostResolveAttempt() throws Exception {
    TaskAttemptImpl.RequestContainerTransition rct = new TaskAttemptImpl.RequestContainerTransition(false);
    EventHandler eventHandler = mock(EventHandler.class);
    String[] hosts = new String[3];
    hosts[0] = "192.168.1.1";
    hosts[1] = "host2";
    hosts[2] = "host3";
    TaskSplitMetaInfo splitInfo = new TaskSplitMetaInfo(hosts, 0, 128 * 1024 * 1024l);
    TaskAttemptImpl mockTaskAttempt = createMapTaskAttemptImplForTest(eventHandler, splitInfo);
    TaskAttemptImpl spyTa = spy(mockTaskAttempt);
    when(spyTa.resolveHost(hosts[0])).thenReturn("host1");
    spyTa.dataLocalHosts = spyTa.resolveHosts(splitInfo.getLocations());
    TaskAttemptEvent mockTAEvent = mock(TaskAttemptEvent.class);
    rct.transition(spyTa, mockTAEvent);
    verify(spyTa).resolveHost(hosts[0]);
    ArgumentCaptor<Event> arg = ArgumentCaptor.forClass(Event.class);
    verify(eventHandler, times(2)).handle(arg.capture());
    if (!(arg.getAllValues().get(1) instanceof ContainerRequestEvent)) {
        Assert.fail("Second Event not of type ContainerRequestEvent");
    }
    Map<String, Boolean> expected = new HashMap<String, Boolean>();
    expected.put("host1", true);
    expected.put("host2", true);
    expected.put("host3", true);
    ContainerRequestEvent cre = (ContainerRequestEvent) arg.getAllValues().get(1);
    String[] requestedHosts = cre.getHosts();
    for (String h : requestedHosts) {
        expected.remove(h);
    }
    assertEquals(0, expected.size());
}
Also used : HashMap(java.util.HashMap) MapTaskAttemptImpl(org.apache.hadoop.mapred.MapTaskAttemptImpl) EventHandler(org.apache.hadoop.yarn.event.EventHandler) TaskAttemptEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptEvent) ContainerRequestEvent(org.apache.hadoop.mapreduce.v2.app.rm.ContainerRequestEvent) TaskAttemptContainerLaunchedEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptContainerLaunchedEvent) TaskTAttemptKilledEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskTAttemptKilledEvent) ContainerRequestEvent(org.apache.hadoop.mapreduce.v2.app.rm.ContainerRequestEvent) Event(org.apache.hadoop.yarn.event.Event) TaskAttemptDiagnosticsUpdateEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptDiagnosticsUpdateEvent) TaskAttemptTooManyFetchFailureEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptTooManyFetchFailureEvent) JobHistoryEvent(org.apache.hadoop.mapreduce.jobhistory.JobHistoryEvent) TaskAttemptContainerAssignedEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptContainerAssignedEvent) JobEvent(org.apache.hadoop.mapreduce.v2.app.job.event.JobEvent) TaskAttemptKillEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptKillEvent) TaskAttemptEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptEvent) TaskEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskEvent) TaskSplitMetaInfo(org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo) Test(org.junit.Test)

Example 5 with TaskSplitMetaInfo

use of org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo in project hadoop by apache.

the class TestTaskAttempt method testContainerCleanedWhileRunning.

@Test
public void testContainerCleanedWhileRunning() throws Exception {
    ApplicationId appId = ApplicationId.newInstance(1, 2);
    ApplicationAttemptId appAttemptId = ApplicationAttemptId.newInstance(appId, 0);
    JobId jobId = MRBuilderUtils.newJobId(appId, 1);
    TaskId taskId = MRBuilderUtils.newTaskId(jobId, 1, TaskType.MAP);
    TaskAttemptId attemptId = MRBuilderUtils.newTaskAttemptId(taskId, 0);
    Path jobFile = mock(Path.class);
    MockEventHandler eventHandler = new MockEventHandler();
    TaskAttemptListener taListener = mock(TaskAttemptListener.class);
    when(taListener.getAddress()).thenReturn(new InetSocketAddress("localhost", 0));
    JobConf jobConf = new JobConf();
    jobConf.setClass("fs.file.impl", StubbedFS.class, FileSystem.class);
    jobConf.setBoolean("fs.file.impl.disable.cache", true);
    jobConf.set(JobConf.MAPRED_MAP_TASK_ENV, "");
    jobConf.set(MRJobConfig.APPLICATION_ATTEMPT_ID, "10");
    TaskSplitMetaInfo splits = mock(TaskSplitMetaInfo.class);
    when(splits.getLocations()).thenReturn(new String[] { "127.0.0.1" });
    AppContext appCtx = mock(AppContext.class);
    ClusterInfo clusterInfo = mock(ClusterInfo.class);
    Resource resource = mock(Resource.class);
    when(appCtx.getClusterInfo()).thenReturn(clusterInfo);
    when(resource.getMemorySize()).thenReturn(1024L);
    setupTaskAttemptFinishingMonitor(eventHandler, jobConf, appCtx);
    TaskAttemptImpl taImpl = new MapTaskAttemptImpl(taskId, 1, eventHandler, jobFile, 1, splits, jobConf, taListener, new Token(), new Credentials(), SystemClock.getInstance(), appCtx);
    NodeId nid = NodeId.newInstance("127.0.0.2", 0);
    ContainerId contId = ContainerId.newContainerId(appAttemptId, 3);
    Container container = mock(Container.class);
    when(container.getId()).thenReturn(contId);
    when(container.getNodeId()).thenReturn(nid);
    when(container.getNodeHttpAddress()).thenReturn("localhost:0");
    taImpl.handle(new TaskAttemptEvent(attemptId, TaskAttemptEventType.TA_SCHEDULE));
    taImpl.handle(new TaskAttemptContainerAssignedEvent(attemptId, container, mock(Map.class)));
    taImpl.handle(new TaskAttemptContainerLaunchedEvent(attemptId, 0));
    assertEquals("Task attempt is not in running state", taImpl.getState(), TaskAttemptState.RUNNING);
    taImpl.handle(new TaskAttemptEvent(attemptId, TaskAttemptEventType.TA_CONTAINER_CLEANED));
    assertFalse("InternalError occurred trying to handle TA_CONTAINER_CLEANED", eventHandler.internalError);
    assertEquals("Task attempt is not assigned on the local rack", Locality.RACK_LOCAL, taImpl.getLocality());
}
Also used : Path(org.apache.hadoop.fs.Path) TaskId(org.apache.hadoop.mapreduce.v2.api.records.TaskId) TaskAttemptId(org.apache.hadoop.mapreduce.v2.api.records.TaskAttemptId) TaskAttemptListener(org.apache.hadoop.mapreduce.v2.app.TaskAttemptListener) InetSocketAddress(java.net.InetSocketAddress) AppContext(org.apache.hadoop.mapreduce.v2.app.AppContext) Resource(org.apache.hadoop.yarn.api.records.Resource) MapTaskAttemptImpl(org.apache.hadoop.mapred.MapTaskAttemptImpl) Token(org.apache.hadoop.security.token.Token) TaskAttemptEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptEvent) ApplicationAttemptId(org.apache.hadoop.yarn.api.records.ApplicationAttemptId) TaskAttemptContainerAssignedEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptContainerAssignedEvent) TaskAttemptContainerLaunchedEvent(org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptContainerLaunchedEvent) ClusterInfo(org.apache.hadoop.mapreduce.v2.app.ClusterInfo) Container(org.apache.hadoop.yarn.api.records.Container) ContainerId(org.apache.hadoop.yarn.api.records.ContainerId) NodeId(org.apache.hadoop.yarn.api.records.NodeId) TaskSplitMetaInfo(org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo) ApplicationId(org.apache.hadoop.yarn.api.records.ApplicationId) MapTaskAttemptImpl(org.apache.hadoop.mapred.MapTaskAttemptImpl) JobConf(org.apache.hadoop.mapred.JobConf) JobId(org.apache.hadoop.mapreduce.v2.api.records.JobId) Credentials(org.apache.hadoop.security.Credentials) Test(org.junit.Test)

Aggregations

TaskSplitMetaInfo (org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo)18 Path (org.apache.hadoop.fs.Path)15 JobId (org.apache.hadoop.mapreduce.v2.api.records.JobId)15 Test (org.junit.Test)15 JobConf (org.apache.hadoop.mapred.JobConf)14 MapTaskAttemptImpl (org.apache.hadoop.mapred.MapTaskAttemptImpl)14 Credentials (org.apache.hadoop.security.Credentials)14 TaskAttemptListener (org.apache.hadoop.mapreduce.v2.app.TaskAttemptListener)13 TaskAttemptEvent (org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptEvent)13 Token (org.apache.hadoop.security.token.Token)13 ApplicationId (org.apache.hadoop.yarn.api.records.ApplicationId)13 InetSocketAddress (java.net.InetSocketAddress)12 TaskAttemptId (org.apache.hadoop.mapreduce.v2.api.records.TaskAttemptId)12 TaskId (org.apache.hadoop.mapreduce.v2.api.records.TaskId)12 AppContext (org.apache.hadoop.mapreduce.v2.app.AppContext)12 TaskAttemptContainerAssignedEvent (org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptContainerAssignedEvent)12 ApplicationAttemptId (org.apache.hadoop.yarn.api.records.ApplicationAttemptId)12 Container (org.apache.hadoop.yarn.api.records.Container)12 ContainerId (org.apache.hadoop.yarn.api.records.ContainerId)12 NodeId (org.apache.hadoop.yarn.api.records.NodeId)12