Search in sources :

Example 31 with TaskAttemptImpl

use of org.apache.hadoop.mapreduce.v2.app.job.impl.TaskAttemptImpl in project hadoop by apache.

the class TestTaskAttempt method testContainerKillAfterAssigned.

@Test
public void testContainerKillAfterAssigned() 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)));
    assertEquals("Task attempt is not in assinged state", taImpl.getInternalState(), TaskAttemptStateInternal.ASSIGNED);
    taImpl.handle(new TaskAttemptEvent(attemptId, TaskAttemptEventType.TA_KILL));
    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) 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 32 with TaskAttemptImpl

use of org.apache.hadoop.mapreduce.v2.app.job.impl.TaskAttemptImpl in project hadoop by apache.

the class TestTaskAttempt method createTaskAttemptImpl.

private TaskAttemptImpl createTaskAttemptImpl(MockEventHandler eventHandler) {
    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);
    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);
    when(appCtx.getClusterInfo()).thenReturn(clusterInfo);
    setupTaskAttemptFinishingMonitor(eventHandler, jobConf, appCtx);
    TaskAttemptImpl taImpl = new MapTaskAttemptImpl(taskId, 1, eventHandler, jobFile, 1, splits, jobConf, taListener, mock(Token.class), new Credentials(), SystemClock.getInstance(), appCtx);
    NodeId nid = NodeId.newInstance("127.0.0.1", 0);
    ContainerId contId = ContainerId.newInstance(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));
    return taImpl;
}
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) 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)

Example 33 with TaskAttemptImpl

use of org.apache.hadoop.mapreduce.v2.app.job.impl.TaskAttemptImpl in project hadoop by apache.

the class TestMRApp method testContainerPassThrough.

@Test
public void testContainerPassThrough() throws Exception {
    MRApp app = new MRApp(0, 1, true, this.getClass().getName(), true) {

        @Override
        protected ContainerLauncher createContainerLauncher(AppContext context) {
            return new MockContainerLauncher() {

                @Override
                public void handle(ContainerLauncherEvent event) {
                    if (event instanceof ContainerRemoteLaunchEvent) {
                        containerObtainedByContainerLauncher = ((ContainerRemoteLaunchEvent) event).getAllocatedContainer();
                    }
                    super.handle(event);
                }
            };
        }

        ;
    };
    Job job = app.submit(new Configuration());
    app.waitForState(job, JobState.SUCCEEDED);
    app.verifyCompleted();
    Collection<Task> tasks = job.getTasks().values();
    Collection<TaskAttempt> taskAttempts = tasks.iterator().next().getAttempts().values();
    TaskAttemptImpl taskAttempt = (TaskAttemptImpl) taskAttempts.iterator().next();
    // Container from RM should pass through to the launcher. Container object
    // should be the same.
    Assert.assertTrue(taskAttempt.container == containerObtainedByContainerLauncher);
}
Also used : Task(org.apache.hadoop.mapreduce.v2.app.job.Task) Configuration(org.apache.hadoop.conf.Configuration) ContainerLauncherEvent(org.apache.hadoop.mapreduce.v2.app.launcher.ContainerLauncherEvent) TaskAttemptImpl(org.apache.hadoop.mapreduce.v2.app.job.impl.TaskAttemptImpl) TaskAttempt(org.apache.hadoop.mapreduce.v2.app.job.TaskAttempt) Job(org.apache.hadoop.mapreduce.v2.app.job.Job) ContainerRemoteLaunchEvent(org.apache.hadoop.mapreduce.v2.app.launcher.ContainerRemoteLaunchEvent) Test(org.junit.Test)

Example 34 with TaskAttemptImpl

use of org.apache.hadoop.mapreduce.v2.app.job.impl.TaskAttemptImpl in project hadoop by apache.

the class TestTaskAttempt method testContainerCleanedWhileCommitting.

@Test
public void testContainerCleanedWhileCommitting() 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[] {});
    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 TaskAttemptEvent(attemptId, TaskAttemptEventType.TA_SCHEDULE));
    taImpl.handle(new TaskAttemptContainerAssignedEvent(attemptId, container, mock(Map.class)));
    taImpl.handle(new TaskAttemptContainerLaunchedEvent(attemptId, 0));
    taImpl.handle(new TaskAttemptEvent(attemptId, TaskAttemptEventType.TA_COMMIT_PENDING));
    assertEquals("Task attempt is not in commit pending state", taImpl.getState(), TaskAttemptState.COMMIT_PENDING);
    taImpl.handle(new TaskAttemptEvent(attemptId, TaskAttemptEventType.TA_CONTAINER_CLEANED));
    assertFalse("InternalError occurred trying to handle TA_CONTAINER_CLEANED", eventHandler.internalError);
    assertEquals("Task attempt is assigned locally", Locality.OFF_SWITCH, 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)

Example 35 with TaskAttemptImpl

use of org.apache.hadoop.mapreduce.v2.app.job.impl.TaskAttemptImpl in project hadoop by apache.

the class TestTaskAttemptContainerRequest method testAttemptContainerRequest.

//WARNING: This test must be the only test in this file.  This is because
// there is an optimization where the credentials passed in are cached
// statically so they do not need to be recomputed when creating a new
// ContainerLaunchContext. if other tests run first this code will cache
// their credentials and this test will fail trying to look for the
// credentials it inserted in.
@Test
public void testAttemptContainerRequest() throws Exception {
    final Text SECRET_KEY_ALIAS = new Text("secretkeyalias");
    final byte[] SECRET_KEY = ("secretkey").getBytes();
    Map<ApplicationAccessType, String> acls = new HashMap<ApplicationAccessType, String>(1);
    acls.put(ApplicationAccessType.VIEW_APP, "otheruser");
    ApplicationId appId = ApplicationId.newInstance(1, 1);
    JobId jobId = MRBuilderUtils.newJobId(appId, 1);
    TaskId taskId = MRBuilderUtils.newTaskId(jobId, 1, TaskType.MAP);
    Path jobFile = mock(Path.class);
    EventHandler eventHandler = mock(EventHandler.class);
    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, "");
    // setup UGI for security so tokens and keys are preserved
    jobConf.set(CommonConfigurationKeysPublic.HADOOP_SECURITY_AUTHENTICATION, "kerberos");
    UserGroupInformation.setConfiguration(jobConf);
    Credentials credentials = new Credentials();
    credentials.addSecretKey(SECRET_KEY_ALIAS, SECRET_KEY);
    Token<JobTokenIdentifier> jobToken = new Token<JobTokenIdentifier>(("tokenid").getBytes(), ("tokenpw").getBytes(), new Text("tokenkind"), new Text("tokenservice"));
    TaskAttemptImpl taImpl = new MapTaskAttemptImpl(taskId, 1, eventHandler, jobFile, 1, mock(TaskSplitMetaInfo.class), jobConf, taListener, jobToken, credentials, SystemClock.getInstance(), null);
    jobConf.set(MRJobConfig.APPLICATION_ATTEMPT_ID, taImpl.getID().toString());
    ContainerLaunchContext launchCtx = TaskAttemptImpl.createContainerLaunchContext(acls, jobConf, jobToken, taImpl.createRemoteTask(), TypeConverter.fromYarn(jobId), mock(WrappedJvmID.class), taListener, credentials);
    Assert.assertEquals("ACLs mismatch", acls, launchCtx.getApplicationACLs());
    Credentials launchCredentials = new Credentials();
    DataInputByteBuffer dibb = new DataInputByteBuffer();
    dibb.reset(launchCtx.getTokens());
    launchCredentials.readTokenStorageStream(dibb);
    // verify all tokens specified for the task attempt are in the launch context
    for (Token<? extends TokenIdentifier> token : credentials.getAllTokens()) {
        Token<? extends TokenIdentifier> launchToken = launchCredentials.getToken(token.getService());
        Assert.assertNotNull("Token " + token.getService() + " is missing", launchToken);
        Assert.assertEquals("Token " + token.getService() + " mismatch", token, launchToken);
    }
    // verify the secret key is in the launch context
    Assert.assertNotNull("Secret key missing", launchCredentials.getSecretKey(SECRET_KEY_ALIAS));
    Assert.assertTrue("Secret key mismatch", Arrays.equals(SECRET_KEY, launchCredentials.getSecretKey(SECRET_KEY_ALIAS)));
}
Also used : TaskId(org.apache.hadoop.mapreduce.v2.api.records.TaskId) HashMap(java.util.HashMap) TaskAttemptListener(org.apache.hadoop.mapreduce.v2.app.TaskAttemptListener) InetSocketAddress(java.net.InetSocketAddress) DataInputByteBuffer(org.apache.hadoop.io.DataInputByteBuffer) EventHandler(org.apache.hadoop.yarn.event.EventHandler) Token(org.apache.hadoop.security.token.Token) WrappedJvmID(org.apache.hadoop.mapred.WrappedJvmID) JobConf(org.apache.hadoop.mapred.JobConf) JobId(org.apache.hadoop.mapreduce.v2.api.records.JobId) Path(org.apache.hadoop.fs.Path) JobTokenIdentifier(org.apache.hadoop.mapreduce.security.token.JobTokenIdentifier) MapTaskAttemptImpl(org.apache.hadoop.mapred.MapTaskAttemptImpl) Text(org.apache.hadoop.io.Text) ContainerLaunchContext(org.apache.hadoop.yarn.api.records.ContainerLaunchContext) ApplicationAccessType(org.apache.hadoop.yarn.api.records.ApplicationAccessType) TaskSplitMetaInfo(org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo) ApplicationId(org.apache.hadoop.yarn.api.records.ApplicationId) MapTaskAttemptImpl(org.apache.hadoop.mapred.MapTaskAttemptImpl) Credentials(org.apache.hadoop.security.Credentials) Test(org.junit.Test)

Aggregations

Test (org.junit.Test)27 MapTaskAttemptImpl (org.apache.hadoop.mapred.MapTaskAttemptImpl)25 TaskAttemptEvent (org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptEvent)22 TaskId (org.apache.hadoop.mapreduce.v2.api.records.TaskId)19 TaskSplitMetaInfo (org.apache.hadoop.mapreduce.split.JobSplit.TaskSplitMetaInfo)17 Path (org.apache.hadoop.fs.Path)16 JobConf (org.apache.hadoop.mapred.JobConf)16 JobId (org.apache.hadoop.mapreduce.v2.api.records.JobId)16 TaskAttemptListener (org.apache.hadoop.mapreduce.v2.app.TaskAttemptListener)16 ApplicationId (org.apache.hadoop.yarn.api.records.ApplicationId)16 InetSocketAddress (java.net.InetSocketAddress)15 Credentials (org.apache.hadoop.security.Credentials)15 Token (org.apache.hadoop.security.token.Token)15 TaskAttemptId (org.apache.hadoop.mapreduce.v2.api.records.TaskAttemptId)14 AppContext (org.apache.hadoop.mapreduce.v2.app.AppContext)14 ApplicationAttemptId (org.apache.hadoop.yarn.api.records.ApplicationAttemptId)14 TaskAttemptContainerAssignedEvent (org.apache.hadoop.mapreduce.v2.app.job.event.TaskAttemptContainerAssignedEvent)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