use of org.apache.hadoop.yarn.exceptions.YarnException in project hadoop by apache.
the class JobHistoryEventHandler method processEventForTimelineServer.
private void processEventForTimelineServer(HistoryEvent event, JobId jobId, long timestamp) {
TimelineEvent tEvent = new TimelineEvent();
tEvent.setEventType(StringUtils.toUpperCase(event.getEventType().name()));
tEvent.setTimestamp(timestamp);
TimelineEntity tEntity = new TimelineEntity();
switch(event.getEventType()) {
case JOB_SUBMITTED:
JobSubmittedEvent jse = (JobSubmittedEvent) event;
tEvent.addEventInfo("SUBMIT_TIME", jse.getSubmitTime());
tEvent.addEventInfo("QUEUE_NAME", jse.getJobQueueName());
tEvent.addEventInfo("JOB_NAME", jse.getJobName());
tEvent.addEventInfo("USER_NAME", jse.getUserName());
tEvent.addEventInfo("JOB_CONF_PATH", jse.getJobConfPath());
tEvent.addEventInfo("ACLS", jse.getJobAcls());
tEvent.addEventInfo("JOB_QUEUE_NAME", jse.getJobQueueName());
tEvent.addEventInfo("WORKFLOW_ID", jse.getWorkflowId());
tEvent.addEventInfo("WORKFLOW_NAME", jse.getWorkflowName());
tEvent.addEventInfo("WORKFLOW_NAME_NAME", jse.getWorkflowNodeName());
tEvent.addEventInfo("WORKFLOW_ADJACENCIES", jse.getWorkflowAdjacencies());
tEvent.addEventInfo("WORKFLOW_TAGS", jse.getWorkflowTags());
tEntity.addEvent(tEvent);
tEntity.setEntityId(jobId.toString());
tEntity.setEntityType(MAPREDUCE_JOB_ENTITY_TYPE);
break;
case JOB_STATUS_CHANGED:
JobStatusChangedEvent jsce = (JobStatusChangedEvent) event;
tEvent.addEventInfo("STATUS", jsce.getStatus());
tEntity.addEvent(tEvent);
tEntity.setEntityId(jobId.toString());
tEntity.setEntityType(MAPREDUCE_JOB_ENTITY_TYPE);
break;
case JOB_INFO_CHANGED:
JobInfoChangeEvent jice = (JobInfoChangeEvent) event;
tEvent.addEventInfo("SUBMIT_TIME", jice.getSubmitTime());
tEvent.addEventInfo("LAUNCH_TIME", jice.getLaunchTime());
tEntity.addEvent(tEvent);
tEntity.setEntityId(jobId.toString());
tEntity.setEntityType(MAPREDUCE_JOB_ENTITY_TYPE);
break;
case JOB_INITED:
JobInitedEvent jie = (JobInitedEvent) event;
tEvent.addEventInfo("START_TIME", jie.getLaunchTime());
tEvent.addEventInfo("STATUS", jie.getStatus());
tEvent.addEventInfo("TOTAL_MAPS", jie.getTotalMaps());
tEvent.addEventInfo("TOTAL_REDUCES", jie.getTotalReduces());
tEvent.addEventInfo("UBERIZED", jie.getUberized());
tEntity.setStartTime(jie.getLaunchTime());
tEntity.addEvent(tEvent);
tEntity.setEntityId(jobId.toString());
tEntity.setEntityType(MAPREDUCE_JOB_ENTITY_TYPE);
break;
case JOB_PRIORITY_CHANGED:
JobPriorityChangeEvent jpce = (JobPriorityChangeEvent) event;
tEvent.addEventInfo("PRIORITY", jpce.getPriority().toString());
tEntity.addEvent(tEvent);
tEntity.setEntityId(jobId.toString());
tEntity.setEntityType(MAPREDUCE_JOB_ENTITY_TYPE);
break;
case JOB_QUEUE_CHANGED:
JobQueueChangeEvent jqe = (JobQueueChangeEvent) event;
tEvent.addEventInfo("QUEUE_NAMES", jqe.getJobQueueName());
tEntity.addEvent(tEvent);
tEntity.setEntityId(jobId.toString());
tEntity.setEntityType(MAPREDUCE_JOB_ENTITY_TYPE);
break;
case JOB_FAILED:
case JOB_KILLED:
case JOB_ERROR:
JobUnsuccessfulCompletionEvent juce = (JobUnsuccessfulCompletionEvent) event;
tEvent.addEventInfo("FINISH_TIME", juce.getFinishTime());
tEvent.addEventInfo("NUM_MAPS", juce.getFinishedMaps());
tEvent.addEventInfo("NUM_REDUCES", juce.getFinishedReduces());
tEvent.addEventInfo("JOB_STATUS", juce.getStatus());
tEvent.addEventInfo("DIAGNOSTICS", juce.getDiagnostics());
tEvent.addEventInfo("FINISHED_MAPS", juce.getFinishedMaps());
tEvent.addEventInfo("FINISHED_REDUCES", juce.getFinishedReduces());
tEntity.addEvent(tEvent);
tEntity.setEntityId(jobId.toString());
tEntity.setEntityType(MAPREDUCE_JOB_ENTITY_TYPE);
break;
case JOB_FINISHED:
JobFinishedEvent jfe = (JobFinishedEvent) event;
tEvent.addEventInfo("FINISH_TIME", jfe.getFinishTime());
tEvent.addEventInfo("NUM_MAPS", jfe.getFinishedMaps());
tEvent.addEventInfo("NUM_REDUCES", jfe.getFinishedReduces());
tEvent.addEventInfo("FAILED_MAPS", jfe.getFailedMaps());
tEvent.addEventInfo("FAILED_REDUCES", jfe.getFailedReduces());
tEvent.addEventInfo("FINISHED_MAPS", jfe.getFinishedMaps());
tEvent.addEventInfo("FINISHED_REDUCES", jfe.getFinishedReduces());
tEvent.addEventInfo("MAP_COUNTERS_GROUPS", JobHistoryEventUtils.countersToJSON(jfe.getMapCounters()));
tEvent.addEventInfo("REDUCE_COUNTERS_GROUPS", JobHistoryEventUtils.countersToJSON(jfe.getReduceCounters()));
tEvent.addEventInfo("TOTAL_COUNTERS_GROUPS", JobHistoryEventUtils.countersToJSON(jfe.getTotalCounters()));
tEvent.addEventInfo("JOB_STATUS", JobState.SUCCEEDED.toString());
tEntity.addEvent(tEvent);
tEntity.setEntityId(jobId.toString());
tEntity.setEntityType(MAPREDUCE_JOB_ENTITY_TYPE);
break;
case TASK_STARTED:
TaskStartedEvent tse = (TaskStartedEvent) event;
tEvent.addEventInfo("TASK_TYPE", tse.getTaskType().toString());
tEvent.addEventInfo("START_TIME", tse.getStartTime());
tEvent.addEventInfo("SPLIT_LOCATIONS", tse.getSplitLocations());
tEntity.addEvent(tEvent);
tEntity.setEntityId(tse.getTaskId().toString());
tEntity.setEntityType(MAPREDUCE_TASK_ENTITY_TYPE);
tEntity.addRelatedEntity(MAPREDUCE_JOB_ENTITY_TYPE, jobId.toString());
break;
case TASK_FAILED:
TaskFailedEvent tfe = (TaskFailedEvent) event;
tEvent.addEventInfo("TASK_TYPE", tfe.getTaskType().toString());
tEvent.addEventInfo("STATUS", TaskStatus.State.FAILED.toString());
tEvent.addEventInfo("FINISH_TIME", tfe.getFinishTime());
tEvent.addEventInfo("ERROR", tfe.getError());
tEvent.addEventInfo("FAILED_ATTEMPT_ID", tfe.getFailedAttemptID() == null ? "" : tfe.getFailedAttemptID().toString());
tEvent.addEventInfo("COUNTERS_GROUPS", JobHistoryEventUtils.countersToJSON(tfe.getCounters()));
tEntity.addEvent(tEvent);
tEntity.setEntityId(tfe.getTaskId().toString());
tEntity.setEntityType(MAPREDUCE_TASK_ENTITY_TYPE);
tEntity.addRelatedEntity(MAPREDUCE_JOB_ENTITY_TYPE, jobId.toString());
break;
case TASK_UPDATED:
TaskUpdatedEvent tue = (TaskUpdatedEvent) event;
tEvent.addEventInfo("FINISH_TIME", tue.getFinishTime());
tEntity.addEvent(tEvent);
tEntity.setEntityId(tue.getTaskId().toString());
tEntity.setEntityType(MAPREDUCE_TASK_ENTITY_TYPE);
tEntity.addRelatedEntity(MAPREDUCE_JOB_ENTITY_TYPE, jobId.toString());
break;
case TASK_FINISHED:
TaskFinishedEvent tfe2 = (TaskFinishedEvent) event;
tEvent.addEventInfo("TASK_TYPE", tfe2.getTaskType().toString());
tEvent.addEventInfo("COUNTERS_GROUPS", JobHistoryEventUtils.countersToJSON(tfe2.getCounters()));
tEvent.addEventInfo("FINISH_TIME", tfe2.getFinishTime());
tEvent.addEventInfo("STATUS", TaskStatus.State.SUCCEEDED.toString());
tEvent.addEventInfo("SUCCESSFUL_TASK_ATTEMPT_ID", tfe2.getSuccessfulTaskAttemptId() == null ? "" : tfe2.getSuccessfulTaskAttemptId().toString());
tEntity.addEvent(tEvent);
tEntity.setEntityId(tfe2.getTaskId().toString());
tEntity.setEntityType(MAPREDUCE_TASK_ENTITY_TYPE);
tEntity.addRelatedEntity(MAPREDUCE_JOB_ENTITY_TYPE, jobId.toString());
break;
case MAP_ATTEMPT_STARTED:
case CLEANUP_ATTEMPT_STARTED:
case REDUCE_ATTEMPT_STARTED:
case SETUP_ATTEMPT_STARTED:
TaskAttemptStartedEvent tase = (TaskAttemptStartedEvent) event;
tEvent.addEventInfo("TASK_TYPE", tase.getTaskType().toString());
tEvent.addEventInfo("TASK_ATTEMPT_ID", tase.getTaskAttemptId().toString());
tEvent.addEventInfo("START_TIME", tase.getStartTime());
tEvent.addEventInfo("HTTP_PORT", tase.getHttpPort());
tEvent.addEventInfo("TRACKER_NAME", tase.getTrackerName());
tEvent.addEventInfo("SHUFFLE_PORT", tase.getShufflePort());
tEvent.addEventInfo("CONTAINER_ID", tase.getContainerId() == null ? "" : tase.getContainerId().toString());
tEntity.addEvent(tEvent);
tEntity.setEntityId(tase.getTaskId().toString());
tEntity.setEntityType(MAPREDUCE_TASK_ENTITY_TYPE);
tEntity.addRelatedEntity(MAPREDUCE_JOB_ENTITY_TYPE, jobId.toString());
break;
case MAP_ATTEMPT_FAILED:
case CLEANUP_ATTEMPT_FAILED:
case REDUCE_ATTEMPT_FAILED:
case SETUP_ATTEMPT_FAILED:
case MAP_ATTEMPT_KILLED:
case CLEANUP_ATTEMPT_KILLED:
case REDUCE_ATTEMPT_KILLED:
case SETUP_ATTEMPT_KILLED:
TaskAttemptUnsuccessfulCompletionEvent tauce = (TaskAttemptUnsuccessfulCompletionEvent) event;
tEvent.addEventInfo("TASK_TYPE", tauce.getTaskType().toString());
tEvent.addEventInfo("TASK_ATTEMPT_ID", tauce.getTaskAttemptId() == null ? "" : tauce.getTaskAttemptId().toString());
tEvent.addEventInfo("FINISH_TIME", tauce.getFinishTime());
tEvent.addEventInfo("ERROR", tauce.getError());
tEvent.addEventInfo("STATUS", tauce.getTaskStatus());
tEvent.addEventInfo("HOSTNAME", tauce.getHostname());
tEvent.addEventInfo("PORT", tauce.getPort());
tEvent.addEventInfo("RACK_NAME", tauce.getRackName());
tEvent.addEventInfo("SHUFFLE_FINISH_TIME", tauce.getFinishTime());
tEvent.addEventInfo("SORT_FINISH_TIME", tauce.getFinishTime());
tEvent.addEventInfo("MAP_FINISH_TIME", tauce.getFinishTime());
tEvent.addEventInfo("COUNTERS_GROUPS", JobHistoryEventUtils.countersToJSON(tauce.getCounters()));
tEntity.addEvent(tEvent);
tEntity.setEntityId(tauce.getTaskId().toString());
tEntity.setEntityType(MAPREDUCE_TASK_ENTITY_TYPE);
tEntity.addRelatedEntity(MAPREDUCE_JOB_ENTITY_TYPE, jobId.toString());
break;
case MAP_ATTEMPT_FINISHED:
MapAttemptFinishedEvent mafe = (MapAttemptFinishedEvent) event;
tEvent.addEventInfo("TASK_TYPE", mafe.getTaskType().toString());
tEvent.addEventInfo("FINISH_TIME", mafe.getFinishTime());
tEvent.addEventInfo("STATUS", mafe.getTaskStatus());
tEvent.addEventInfo("STATE", mafe.getState());
tEvent.addEventInfo("MAP_FINISH_TIME", mafe.getMapFinishTime());
tEvent.addEventInfo("COUNTERS_GROUPS", JobHistoryEventUtils.countersToJSON(mafe.getCounters()));
tEvent.addEventInfo("HOSTNAME", mafe.getHostname());
tEvent.addEventInfo("PORT", mafe.getPort());
tEvent.addEventInfo("RACK_NAME", mafe.getRackName());
tEvent.addEventInfo("ATTEMPT_ID", mafe.getAttemptId() == null ? "" : mafe.getAttemptId().toString());
tEntity.addEvent(tEvent);
tEntity.setEntityId(mafe.getTaskId().toString());
tEntity.setEntityType(MAPREDUCE_TASK_ENTITY_TYPE);
tEntity.addRelatedEntity(MAPREDUCE_JOB_ENTITY_TYPE, jobId.toString());
break;
case REDUCE_ATTEMPT_FINISHED:
ReduceAttemptFinishedEvent rafe = (ReduceAttemptFinishedEvent) event;
tEvent.addEventInfo("TASK_TYPE", rafe.getTaskType().toString());
tEvent.addEventInfo("ATTEMPT_ID", rafe.getAttemptId() == null ? "" : rafe.getAttemptId().toString());
tEvent.addEventInfo("FINISH_TIME", rafe.getFinishTime());
tEvent.addEventInfo("STATUS", rafe.getTaskStatus());
tEvent.addEventInfo("STATE", rafe.getState());
tEvent.addEventInfo("SHUFFLE_FINISH_TIME", rafe.getShuffleFinishTime());
tEvent.addEventInfo("SORT_FINISH_TIME", rafe.getSortFinishTime());
tEvent.addEventInfo("COUNTERS_GROUPS", JobHistoryEventUtils.countersToJSON(rafe.getCounters()));
tEvent.addEventInfo("HOSTNAME", rafe.getHostname());
tEvent.addEventInfo("PORT", rafe.getPort());
tEvent.addEventInfo("RACK_NAME", rafe.getRackName());
tEntity.addEvent(tEvent);
tEntity.setEntityId(rafe.getTaskId().toString());
tEntity.setEntityType(MAPREDUCE_TASK_ENTITY_TYPE);
tEntity.addRelatedEntity(MAPREDUCE_JOB_ENTITY_TYPE, jobId.toString());
break;
case SETUP_ATTEMPT_FINISHED:
case CLEANUP_ATTEMPT_FINISHED:
TaskAttemptFinishedEvent tafe = (TaskAttemptFinishedEvent) event;
tEvent.addEventInfo("TASK_TYPE", tafe.getTaskType().toString());
tEvent.addEventInfo("ATTEMPT_ID", tafe.getAttemptId() == null ? "" : tafe.getAttemptId().toString());
tEvent.addEventInfo("FINISH_TIME", tafe.getFinishTime());
tEvent.addEventInfo("STATUS", tafe.getTaskStatus());
tEvent.addEventInfo("STATE", tafe.getState());
tEvent.addEventInfo("COUNTERS_GROUPS", JobHistoryEventUtils.countersToJSON(tafe.getCounters()));
tEvent.addEventInfo("HOSTNAME", tafe.getHostname());
tEntity.addEvent(tEvent);
tEntity.setEntityId(tafe.getTaskId().toString());
tEntity.setEntityType(MAPREDUCE_TASK_ENTITY_TYPE);
tEntity.addRelatedEntity(MAPREDUCE_JOB_ENTITY_TYPE, jobId.toString());
break;
case AM_STARTED:
AMStartedEvent ase = (AMStartedEvent) event;
tEvent.addEventInfo("APPLICATION_ATTEMPT_ID", ase.getAppAttemptId() == null ? "" : ase.getAppAttemptId().toString());
tEvent.addEventInfo("CONTAINER_ID", ase.getContainerId() == null ? "" : ase.getContainerId().toString());
tEvent.addEventInfo("NODE_MANAGER_HOST", ase.getNodeManagerHost());
tEvent.addEventInfo("NODE_MANAGER_PORT", ase.getNodeManagerPort());
tEvent.addEventInfo("NODE_MANAGER_HTTP_PORT", ase.getNodeManagerHttpPort());
tEvent.addEventInfo("START_TIME", ase.getStartTime());
tEvent.addEventInfo("SUBMIT_TIME", ase.getSubmitTime());
tEntity.addEvent(tEvent);
tEntity.setEntityId(jobId.toString());
tEntity.setEntityType(MAPREDUCE_JOB_ENTITY_TYPE);
break;
default:
break;
}
try {
TimelinePutResponse response = timelineClient.putEntities(tEntity);
List<TimelinePutResponse.TimelinePutError> errors = response.getErrors();
if (errors.size() == 0) {
if (LOG.isDebugEnabled()) {
LOG.debug("Timeline entities are successfully put in event " + event.getEventType());
}
} else {
for (TimelinePutResponse.TimelinePutError error : errors) {
LOG.error("Error when publishing entity [" + error.getEntityType() + "," + error.getEntityId() + "], server side error code: " + error.getErrorCode());
}
}
} catch (YarnException | IOException | ClientHandlerException ex) {
LOG.error("Error putting entity " + tEntity.getEntityId() + " to Timeline" + "Server", ex);
}
}
use of org.apache.hadoop.yarn.exceptions.YarnException in project hadoop by apache.
the class TestClientServiceDelegate method testRMDownRestoreForJobStatusBeforeGetAMReport.
@Test
public void testRMDownRestoreForJobStatusBeforeGetAMReport() throws IOException {
Configuration conf = new YarnConfiguration();
conf.setInt(MRJobConfig.MR_CLIENT_MAX_RETRIES, 3);
conf.set(MRConfig.FRAMEWORK_NAME, MRConfig.YARN_FRAMEWORK_NAME);
conf.setBoolean(MRJobConfig.JOB_AM_ACCESS_DISABLED, !isAMReachableFromClient);
MRClientProtocol historyServerProxy = mock(MRClientProtocol.class);
when(historyServerProxy.getJobReport(any(GetJobReportRequest.class))).thenReturn(getJobReportResponse());
ResourceMgrDelegate rmDelegate = mock(ResourceMgrDelegate.class);
try {
when(rmDelegate.getApplicationReport(jobId.getAppId())).thenThrow(new java.lang.reflect.UndeclaredThrowableException(new IOException("Connection refuced1"))).thenThrow(new java.lang.reflect.UndeclaredThrowableException(new IOException("Connection refuced2"))).thenReturn(getFinishedApplicationReport());
ClientServiceDelegate clientServiceDelegate = new ClientServiceDelegate(conf, rmDelegate, oldJobId, historyServerProxy);
JobStatus jobStatus = clientServiceDelegate.getJobStatus(oldJobId);
verify(rmDelegate, times(3)).getApplicationReport(any(ApplicationId.class));
Assert.assertNotNull(jobStatus);
} catch (YarnException e) {
throw new IOException(e);
}
}
use of org.apache.hadoop.yarn.exceptions.YarnException in project hadoop by apache.
the class TestClientServiceDelegate method testReconnectOnAMRestart.
@Test
public void testReconnectOnAMRestart() throws IOException {
//as instantiateAMProxy is not called at all
if (!isAMReachableFromClient) {
return;
}
MRClientProtocol historyServerProxy = mock(MRClientProtocol.class);
// RM returns AM1 url, null, null and AM2 url on invocations.
// Nulls simulate the time when AM2 is in the process of restarting.
ResourceMgrDelegate rmDelegate = mock(ResourceMgrDelegate.class);
try {
when(rmDelegate.getApplicationReport(jobId.getAppId())).thenReturn(getRunningApplicationReport("am1", 78)).thenReturn(getRunningApplicationReport(null, 0)).thenReturn(getRunningApplicationReport(null, 0)).thenReturn(getRunningApplicationReport("am2", 90));
} catch (YarnException e) {
throw new IOException(e);
}
GetJobReportResponse jobReportResponse1 = mock(GetJobReportResponse.class);
when(jobReportResponse1.getJobReport()).thenReturn(MRBuilderUtils.newJobReport(jobId, "jobName-firstGen", "user", JobState.RUNNING, 0, 0, 0, 0, 0, 0, 0, "anything", null, false, ""));
// First AM returns a report with jobName firstGen and simulates AM shutdown
// on second invocation.
MRClientProtocol firstGenAMProxy = mock(MRClientProtocol.class);
when(firstGenAMProxy.getJobReport(any(GetJobReportRequest.class))).thenReturn(jobReportResponse1).thenThrow(new RuntimeException("AM is down!"));
GetJobReportResponse jobReportResponse2 = mock(GetJobReportResponse.class);
when(jobReportResponse2.getJobReport()).thenReturn(MRBuilderUtils.newJobReport(jobId, "jobName-secondGen", "user", JobState.RUNNING, 0, 0, 0, 0, 0, 0, 0, "anything", null, false, ""));
// Second AM generation returns a report with jobName secondGen
MRClientProtocol secondGenAMProxy = mock(MRClientProtocol.class);
when(secondGenAMProxy.getJobReport(any(GetJobReportRequest.class))).thenReturn(jobReportResponse2);
ClientServiceDelegate clientServiceDelegate = spy(getClientServiceDelegate(historyServerProxy, rmDelegate));
// First time, connection should be to AM1, then to AM2. Further requests
// should use the same proxy to AM2 and so instantiateProxy shouldn't be
// called.
doReturn(firstGenAMProxy).doReturn(secondGenAMProxy).when(clientServiceDelegate).instantiateAMProxy(any(InetSocketAddress.class));
JobStatus jobStatus = clientServiceDelegate.getJobStatus(oldJobId);
Assert.assertNotNull(jobStatus);
Assert.assertEquals("jobName-firstGen", jobStatus.getJobName());
jobStatus = clientServiceDelegate.getJobStatus(oldJobId);
Assert.assertNotNull(jobStatus);
Assert.assertEquals("jobName-secondGen", jobStatus.getJobName());
jobStatus = clientServiceDelegate.getJobStatus(oldJobId);
Assert.assertNotNull(jobStatus);
Assert.assertEquals("jobName-secondGen", jobStatus.getJobName());
verify(clientServiceDelegate, times(2)).instantiateAMProxy(any(InetSocketAddress.class));
}
use of org.apache.hadoop.yarn.exceptions.YarnException in project hadoop by apache.
the class BaseAMRMProxyE2ETest method createAMRMProtocol.
protected ApplicationMasterProtocol createAMRMProtocol(YarnClient rmClient, ApplicationId appId, MiniYARNCluster cluster, final Configuration yarnConf) throws IOException, InterruptedException, YarnException {
UserGroupInformation user = null;
// Get the AMRMToken from AMRMProxy
ApplicationReport report = rmClient.getApplicationReport(appId);
user = UserGroupInformation.createProxyUser(report.getCurrentApplicationAttemptId().toString(), UserGroupInformation.getCurrentUser());
ContainerManagerImpl containerManager = (ContainerManagerImpl) cluster.getNodeManager(0).getNMContext().getContainerManager();
AMRMProxyTokenSecretManager amrmTokenSecretManager = containerManager.getAMRMProxyService().getSecretManager();
org.apache.hadoop.security.token.Token<AMRMTokenIdentifier> token = amrmTokenSecretManager.createAndGetAMRMToken(report.getCurrentApplicationAttemptId());
SecurityUtil.setTokenService(token, containerManager.getAMRMProxyService().getBindAddress());
user.addToken(token);
return user.doAs(new PrivilegedExceptionAction<ApplicationMasterProtocol>() {
@Override
public ApplicationMasterProtocol run() throws Exception {
return ClientRMProxy.createRMProxy(yarnConf, ApplicationMasterProtocol.class);
}
});
}
use of org.apache.hadoop.yarn.exceptions.YarnException in project hadoop by apache.
the class TestAMRMClient method allocateAndStartContainers.
private List<Container> allocateAndStartContainers(final AMRMClient<ContainerRequest> amClient, final NMClient nmClient, int num) throws YarnException, IOException {
// set up allocation requests
for (int i = 0; i < num; ++i) {
amClient.addContainerRequest(new ContainerRequest(capability, nodes, racks, priority));
}
// send allocation requests
amClient.allocate(0.1f);
// let NM heartbeat to RM and trigger allocations
triggerSchedulingWithNMHeartBeat();
// get allocations
AllocateResponse allocResponse = amClient.allocate(0.1f);
List<Container> containers = allocResponse.getAllocatedContainers();
Assert.assertEquals(num, containers.size());
// build container launch context
Credentials ts = new Credentials();
DataOutputBuffer dob = new DataOutputBuffer();
ts.writeTokenStorageToStream(dob);
ByteBuffer securityTokens = ByteBuffer.wrap(dob.getData(), 0, dob.getLength());
// start a process long enough for increase/decrease action to take effect
ContainerLaunchContext clc = BuilderUtils.newContainerLaunchContext(Collections.<String, LocalResource>emptyMap(), new HashMap<String, String>(), Arrays.asList("sleep", "100"), new HashMap<String, ByteBuffer>(), securityTokens, new HashMap<ApplicationAccessType, String>());
// start the containers and make sure they are in RUNNING state
try {
for (int i = 0; i < num; i++) {
Container container = containers.get(i);
nmClient.startContainer(container, clc);
// container status
while (true) {
ContainerStatus status = nmClient.getContainerStatus(container.getId(), container.getNodeId());
if (status.getState() == ContainerState.RUNNING) {
break;
}
sleep(10);
}
}
} catch (YarnException e) {
throw new AssertionError("Exception is not expected: " + e);
}
// let NM's heartbeat to RM to confirm container launch
triggerSchedulingWithNMHeartBeat();
return containers;
}
Aggregations