Search in sources :

Example 1 with EventReader

use of org.apache.hadoop.mapreduce.jobhistory.EventReader in project hadoop by apache.

the class MRAppMaster method readJustAMInfos.

private List<AMInfo> readJustAMInfos() {
    List<AMInfo> amInfos = new ArrayList<AMInfo>();
    FSDataInputStream inputStream = null;
    try {
        inputStream = getPreviousJobHistoryStream(getConfig(), appAttemptID);
        EventReader jobHistoryEventReader = new EventReader(inputStream);
        // All AMInfos are contiguous. Track when the first AMStartedEvent
        // appears.
        boolean amStartedEventsBegan = false;
        HistoryEvent event;
        while ((event = jobHistoryEventReader.getNextEvent()) != null) {
            if (event.getEventType() == EventType.AM_STARTED) {
                if (!amStartedEventsBegan) {
                    // First AMStartedEvent.
                    amStartedEventsBegan = true;
                }
                AMStartedEvent amStartedEvent = (AMStartedEvent) event;
                amInfos.add(MRBuilderUtils.newAMInfo(amStartedEvent.getAppAttemptId(), amStartedEvent.getStartTime(), amStartedEvent.getContainerId(), StringInterner.weakIntern(amStartedEvent.getNodeManagerHost()), amStartedEvent.getNodeManagerPort(), amStartedEvent.getNodeManagerHttpPort()));
            } else if (amStartedEventsBegan) {
                // No need to continue reading all the other events.
                break;
            }
        }
    } catch (IOException e) {
        LOG.warn("Could not parse the old history file. " + "Will not have old AMinfos ", e);
    } finally {
        if (inputStream != null) {
            IOUtils.closeQuietly(inputStream);
        }
    }
    return amInfos;
}
Also used : AMInfo(org.apache.hadoop.mapreduce.v2.api.records.AMInfo) AMStartedEvent(org.apache.hadoop.mapreduce.jobhistory.AMStartedEvent) EventReader(org.apache.hadoop.mapreduce.jobhistory.EventReader) ArrayList(java.util.ArrayList) FSDataInputStream(org.apache.hadoop.fs.FSDataInputStream) IOException(java.io.IOException) JobHistoryEvent(org.apache.hadoop.mapreduce.jobhistory.JobHistoryEvent) HistoryEvent(org.apache.hadoop.mapreduce.jobhistory.HistoryEvent)

Example 2 with EventReader

use of org.apache.hadoop.mapreduce.jobhistory.EventReader in project hadoop by apache.

the class TestJobHistoryParsing method checkHistoryParsing.

private void checkHistoryParsing(final int numMaps, final int numReduces, final int numSuccessfulMaps) throws Exception {
    Configuration conf = new Configuration();
    conf.set(MRJobConfig.USER_NAME, System.getProperty("user.name"));
    long amStartTimeEst = System.currentTimeMillis();
    conf.setClass(NET_TOPOLOGY_NODE_SWITCH_MAPPING_IMPL_KEY, MyResolver.class, DNSToSwitchMapping.class);
    RackResolver.init(conf);
    MRApp app = new MRAppWithHistory(numMaps, numReduces, true, this.getClass().getName(), true);
    app.submit(conf);
    Job job = app.getContext().getAllJobs().values().iterator().next();
    JobId jobId = job.getID();
    LOG.info("JOBID is " + TypeConverter.fromYarn(jobId).toString());
    app.waitForState(job, JobState.SUCCEEDED);
    // make sure all events are flushed
    app.waitForState(Service.STATE.STOPPED);
    String jobhistoryDir = JobHistoryUtils.getHistoryIntermediateDoneDirForUser(conf);
    FileContext fc = null;
    try {
        fc = FileContext.getFileContext(conf);
    } catch (IOException ioe) {
        LOG.info("Can not get FileContext", ioe);
        throw (new Exception("Can not get File Context"));
    }
    if (numMaps == numSuccessfulMaps) {
        String summaryFileName = JobHistoryUtils.getIntermediateSummaryFileName(jobId);
        Path summaryFile = new Path(jobhistoryDir, summaryFileName);
        String jobSummaryString = getJobSummary(fc, summaryFile);
        Assert.assertNotNull(jobSummaryString);
        Assert.assertTrue(jobSummaryString.contains("resourcesPerMap=100"));
        Assert.assertTrue(jobSummaryString.contains("resourcesPerReduce=100"));
        Map<String, String> jobSummaryElements = new HashMap<String, String>();
        StringTokenizer strToken = new StringTokenizer(jobSummaryString, ",");
        while (strToken.hasMoreTokens()) {
            String keypair = strToken.nextToken();
            jobSummaryElements.put(keypair.split("=")[0], keypair.split("=")[1]);
        }
        Assert.assertEquals("JobId does not match", jobId.toString(), jobSummaryElements.get("jobId"));
        Assert.assertEquals("JobName does not match", "test", jobSummaryElements.get("jobName"));
        Assert.assertTrue("submitTime should not be 0", Long.parseLong(jobSummaryElements.get("submitTime")) != 0);
        Assert.assertTrue("launchTime should not be 0", Long.parseLong(jobSummaryElements.get("launchTime")) != 0);
        Assert.assertTrue("firstMapTaskLaunchTime should not be 0", Long.parseLong(jobSummaryElements.get("firstMapTaskLaunchTime")) != 0);
        Assert.assertTrue("firstReduceTaskLaunchTime should not be 0", Long.parseLong(jobSummaryElements.get("firstReduceTaskLaunchTime")) != 0);
        Assert.assertTrue("finishTime should not be 0", Long.parseLong(jobSummaryElements.get("finishTime")) != 0);
        Assert.assertEquals("Mismatch in num map slots", numSuccessfulMaps, Integer.parseInt(jobSummaryElements.get("numMaps")));
        Assert.assertEquals("Mismatch in num reduce slots", numReduces, Integer.parseInt(jobSummaryElements.get("numReduces")));
        Assert.assertEquals("User does not match", System.getProperty("user.name"), jobSummaryElements.get("user"));
        Assert.assertEquals("Queue does not match", "default", jobSummaryElements.get("queue"));
        Assert.assertEquals("Status does not match", "SUCCEEDED", jobSummaryElements.get("status"));
    }
    JobHistory jobHistory = new JobHistory();
    jobHistory.init(conf);
    HistoryFileInfo fileInfo = jobHistory.getJobFileInfo(jobId);
    JobInfo jobInfo;
    long numFinishedMaps;
    synchronized (fileInfo) {
        Path historyFilePath = fileInfo.getHistoryFile();
        FSDataInputStream in = null;
        LOG.info("JobHistoryFile is: " + historyFilePath);
        try {
            in = fc.open(fc.makeQualified(historyFilePath));
        } catch (IOException ioe) {
            LOG.info("Can not open history file: " + historyFilePath, ioe);
            throw (new Exception("Can not open History File"));
        }
        JobHistoryParser parser = new JobHistoryParser(in);
        final EventReader realReader = new EventReader(in);
        EventReader reader = Mockito.mock(EventReader.class);
        if (numMaps == numSuccessfulMaps) {
            reader = realReader;
        } else {
            // Hack!
            final AtomicInteger numFinishedEvents = new AtomicInteger(0);
            Mockito.when(reader.getNextEvent()).thenAnswer(new Answer<HistoryEvent>() {

                public HistoryEvent answer(InvocationOnMock invocation) throws IOException {
                    HistoryEvent event = realReader.getNextEvent();
                    if (event instanceof TaskFinishedEvent) {
                        numFinishedEvents.incrementAndGet();
                    }
                    if (numFinishedEvents.get() <= numSuccessfulMaps) {
                        return event;
                    } else {
                        throw new IOException("test");
                    }
                }
            });
        }
        jobInfo = parser.parse(reader);
        numFinishedMaps = computeFinishedMaps(jobInfo, numMaps, numSuccessfulMaps);
        if (numFinishedMaps != numMaps) {
            Exception parseException = parser.getParseException();
            Assert.assertNotNull("Didn't get expected parse exception", parseException);
        }
    }
    Assert.assertEquals("Incorrect username ", System.getProperty("user.name"), jobInfo.getUsername());
    Assert.assertEquals("Incorrect jobName ", "test", jobInfo.getJobname());
    Assert.assertEquals("Incorrect queuename ", "default", jobInfo.getJobQueueName());
    Assert.assertEquals("incorrect conf path", "test", jobInfo.getJobConfPath());
    Assert.assertEquals("incorrect finishedMap ", numSuccessfulMaps, numFinishedMaps);
    Assert.assertEquals("incorrect finishedReduces ", numReduces, jobInfo.getFinishedReduces());
    Assert.assertEquals("incorrect uberized ", job.isUber(), jobInfo.getUberized());
    Map<TaskID, TaskInfo> allTasks = jobInfo.getAllTasks();
    int totalTasks = allTasks.size();
    Assert.assertEquals("total number of tasks is incorrect  ", (numMaps + numReduces), totalTasks);
    // Verify aminfo
    Assert.assertEquals(1, jobInfo.getAMInfos().size());
    Assert.assertEquals(MRApp.NM_HOST, jobInfo.getAMInfos().get(0).getNodeManagerHost());
    AMInfo amInfo = jobInfo.getAMInfos().get(0);
    Assert.assertEquals(MRApp.NM_PORT, amInfo.getNodeManagerPort());
    Assert.assertEquals(MRApp.NM_HTTP_PORT, amInfo.getNodeManagerHttpPort());
    Assert.assertEquals(1, amInfo.getAppAttemptId().getAttemptId());
    Assert.assertEquals(amInfo.getAppAttemptId(), amInfo.getContainerId().getApplicationAttemptId());
    Assert.assertTrue(amInfo.getStartTime() <= System.currentTimeMillis() && amInfo.getStartTime() >= amStartTimeEst);
    ContainerId fakeCid = MRApp.newContainerId(-1, -1, -1, -1);
    // Assert at taskAttempt level
    for (TaskInfo taskInfo : allTasks.values()) {
        int taskAttemptCount = taskInfo.getAllTaskAttempts().size();
        Assert.assertEquals("total number of task attempts ", 1, taskAttemptCount);
        TaskAttemptInfo taInfo = taskInfo.getAllTaskAttempts().values().iterator().next();
        Assert.assertNotNull(taInfo.getContainerId());
        // Verify the wrong ctor is not being used. Remove after mrv1 is removed.
        Assert.assertFalse(taInfo.getContainerId().equals(fakeCid));
    }
    // Deep compare Job and JobInfo
    for (Task task : job.getTasks().values()) {
        TaskInfo taskInfo = allTasks.get(TypeConverter.fromYarn(task.getID()));
        Assert.assertNotNull("TaskInfo not found", taskInfo);
        for (TaskAttempt taskAttempt : task.getAttempts().values()) {
            TaskAttemptInfo taskAttemptInfo = taskInfo.getAllTaskAttempts().get(TypeConverter.fromYarn((taskAttempt.getID())));
            Assert.assertNotNull("TaskAttemptInfo not found", taskAttemptInfo);
            Assert.assertEquals("Incorrect shuffle port for task attempt", taskAttempt.getShufflePort(), taskAttemptInfo.getShufflePort());
            if (numMaps == numSuccessfulMaps) {
                Assert.assertEquals(MRApp.NM_HOST, taskAttemptInfo.getHostname());
                Assert.assertEquals(MRApp.NM_PORT, taskAttemptInfo.getPort());
                // Verify rack-name
                Assert.assertEquals("rack-name is incorrect", taskAttemptInfo.getRackname(), RACK_NAME);
            }
        }
    }
    // test output for HistoryViewer
    PrintStream stdps = System.out;
    try {
        System.setOut(new PrintStream(outContent));
        HistoryViewer viewer;
        synchronized (fileInfo) {
            viewer = new HistoryViewer(fc.makeQualified(fileInfo.getHistoryFile()).toString(), conf, true);
        }
        viewer.print();
        for (TaskInfo taskInfo : allTasks.values()) {
            String test = (taskInfo.getTaskStatus() == null ? "" : taskInfo.getTaskStatus()) + " " + taskInfo.getTaskType() + " task list for " + taskInfo.getTaskId().getJobID();
            Assert.assertTrue(outContent.toString().indexOf(test) > 0);
            Assert.assertTrue(outContent.toString().indexOf(taskInfo.getTaskId().toString()) > 0);
        }
    } finally {
        System.setOut(stdps);
    }
}
Also used : MRAppWithHistory(org.apache.hadoop.mapreduce.v2.hs.TestJobHistoryEvents.MRAppWithHistory) Task(org.apache.hadoop.mapreduce.v2.app.job.Task) Configuration(org.apache.hadoop.conf.Configuration) HashMap(java.util.HashMap) TaskInfo(org.apache.hadoop.mapreduce.jobhistory.JobHistoryParser.TaskInfo) HistoryViewer(org.apache.hadoop.mapreduce.jobhistory.HistoryViewer) JobInfo(org.apache.hadoop.mapreduce.jobhistory.JobHistoryParser.JobInfo) ContainerId(org.apache.hadoop.yarn.api.records.ContainerId) TaskAttempt(org.apache.hadoop.mapreduce.v2.app.job.TaskAttempt) Job(org.apache.hadoop.mapreduce.v2.app.job.Job) JobId(org.apache.hadoop.mapreduce.v2.api.records.JobId) MRApp(org.apache.hadoop.mapreduce.v2.app.MRApp) Path(org.apache.hadoop.fs.Path) EventReader(org.apache.hadoop.mapreduce.jobhistory.EventReader) PrintStream(java.io.PrintStream) HistoryFileInfo(org.apache.hadoop.mapreduce.v2.hs.HistoryFileManager.HistoryFileInfo) TaskID(org.apache.hadoop.mapreduce.TaskID) IOException(java.io.IOException) HistoryEvent(org.apache.hadoop.mapreduce.jobhistory.HistoryEvent) IOException(java.io.IOException) AMInfo(org.apache.hadoop.mapreduce.jobhistory.JobHistoryParser.AMInfo) StringTokenizer(java.util.StringTokenizer) TaskFinishedEvent(org.apache.hadoop.mapreduce.jobhistory.TaskFinishedEvent) JobHistoryParser(org.apache.hadoop.mapreduce.jobhistory.JobHistoryParser) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) InvocationOnMock(org.mockito.invocation.InvocationOnMock) TaskAttemptInfo(org.apache.hadoop.mapreduce.jobhistory.JobHistoryParser.TaskAttemptInfo) FSDataInputStream(org.apache.hadoop.fs.FSDataInputStream) FileContext(org.apache.hadoop.fs.FileContext)

Example 3 with EventReader

use of org.apache.hadoop.mapreduce.jobhistory.EventReader in project hadoop by apache.

the class TestJobHistoryParsing method testMultipleFailedTasks.

@Test
public void testMultipleFailedTasks() throws Exception {
    JobHistoryParser parser = new JobHistoryParser(Mockito.mock(FSDataInputStream.class));
    EventReader reader = Mockito.mock(EventReader.class);
    // Hack!
    final AtomicInteger numEventsRead = new AtomicInteger(0);
    final org.apache.hadoop.mapreduce.TaskType taskType = org.apache.hadoop.mapreduce.TaskType.MAP;
    final TaskID[] tids = new TaskID[2];
    final JobID jid = new JobID("1", 1);
    tids[0] = new TaskID(jid, taskType, 0);
    tids[1] = new TaskID(jid, taskType, 1);
    Mockito.when(reader.getNextEvent()).thenAnswer(new Answer<HistoryEvent>() {

        public HistoryEvent answer(InvocationOnMock invocation) throws IOException {
            // send two task start and two task fail events for tasks 0 and 1
            int eventId = numEventsRead.getAndIncrement();
            TaskID tid = tids[eventId & 0x1];
            if (eventId < 2) {
                return new TaskStartedEvent(tid, 0, taskType, "");
            }
            if (eventId < 4) {
                TaskFailedEvent tfe = new TaskFailedEvent(tid, 0, taskType, "failed", "FAILED", null, new Counters());
                tfe.setDatum(tfe.getDatum());
                return tfe;
            }
            if (eventId < 5) {
                JobUnsuccessfulCompletionEvent juce = new JobUnsuccessfulCompletionEvent(jid, 100L, 2, 0, "JOB_FAILED", Collections.singletonList("Task failed: " + tids[0].toString()));
                return juce;
            }
            return null;
        }
    });
    JobInfo info = parser.parse(reader);
    assertTrue("Task 0 not implicated", info.getErrorInfo().contains(tids[0].toString()));
}
Also used : EventReader(org.apache.hadoop.mapreduce.jobhistory.EventReader) TaskID(org.apache.hadoop.mapreduce.TaskID) JobUnsuccessfulCompletionEvent(org.apache.hadoop.mapreduce.jobhistory.JobUnsuccessfulCompletionEvent) IOException(java.io.IOException) HistoryEvent(org.apache.hadoop.mapreduce.jobhistory.HistoryEvent) TaskStartedEvent(org.apache.hadoop.mapreduce.jobhistory.TaskStartedEvent) JobHistoryParser(org.apache.hadoop.mapreduce.jobhistory.JobHistoryParser) AtomicInteger(java.util.concurrent.atomic.AtomicInteger) JobInfo(org.apache.hadoop.mapreduce.jobhistory.JobHistoryParser.JobInfo) InvocationOnMock(org.mockito.invocation.InvocationOnMock) TaskFailedEvent(org.apache.hadoop.mapreduce.jobhistory.TaskFailedEvent) FSDataInputStream(org.apache.hadoop.fs.FSDataInputStream) Counters(org.apache.hadoop.mapreduce.Counters) JobID(org.apache.hadoop.mapreduce.JobID) Test(org.junit.Test)

Aggregations

IOException (java.io.IOException)3 FSDataInputStream (org.apache.hadoop.fs.FSDataInputStream)3 EventReader (org.apache.hadoop.mapreduce.jobhistory.EventReader)3 HistoryEvent (org.apache.hadoop.mapreduce.jobhistory.HistoryEvent)3 AtomicInteger (java.util.concurrent.atomic.AtomicInteger)2 TaskID (org.apache.hadoop.mapreduce.TaskID)2 JobHistoryParser (org.apache.hadoop.mapreduce.jobhistory.JobHistoryParser)2 JobInfo (org.apache.hadoop.mapreduce.jobhistory.JobHistoryParser.JobInfo)2 InvocationOnMock (org.mockito.invocation.InvocationOnMock)2 PrintStream (java.io.PrintStream)1 ArrayList (java.util.ArrayList)1 HashMap (java.util.HashMap)1 StringTokenizer (java.util.StringTokenizer)1 Configuration (org.apache.hadoop.conf.Configuration)1 FileContext (org.apache.hadoop.fs.FileContext)1 Path (org.apache.hadoop.fs.Path)1 Counters (org.apache.hadoop.mapreduce.Counters)1 JobID (org.apache.hadoop.mapreduce.JobID)1 AMStartedEvent (org.apache.hadoop.mapreduce.jobhistory.AMStartedEvent)1 HistoryViewer (org.apache.hadoop.mapreduce.jobhistory.HistoryViewer)1