/// <exception cref="Org.Apache.Hadoop.Yarn.Exceptions.YarnException"/>
private RMAppImpl CreateAndPopulateNewRMApp(ApplicationSubmissionContext submissionContext
, long submitTime, string user, bool isRecovery)
{
ApplicationId applicationId = submissionContext.GetApplicationId();
ResourceRequest amReq = ValidateAndCreateResourceRequest(submissionContext, isRecovery
);
// Create RMApp
RMAppImpl application = new RMAppImpl(applicationId, rmContext, this.conf, submissionContext
.GetApplicationName(), user, submissionContext.GetQueue(), submissionContext, this
.scheduler, this.masterService, submitTime, submissionContext.GetApplicationType
(), submissionContext.GetApplicationTags(), amReq);
// Concurrent app submissions with same applicationId will fail here
// Concurrent app submissions with different applicationIds will not
// influence each other
if (rmContext.GetRMApps().PutIfAbsent(applicationId, application) != null)
{
string message = "Application with id " + applicationId + " is already present! Cannot add a duplicate!";
Log.Warn(message);
throw new YarnException(message);
}
// Inform the ACLs Manager
this.applicationACLsManager.AddApplication(applicationId, submissionContext.GetAMContainerSpec
().GetApplicationACLs());
string appViewACLs = submissionContext.GetAMContainerSpec().GetApplicationACLs()[
ApplicationAccessType.ViewApp];
rmContext.GetSystemMetricsPublisher().AppACLsUpdated(application, appViewACLs, Runtime
.CurrentTimeMillis());
return(application);
}
/// <exception cref="Org.Apache.Hadoop.Yarn.Exceptions.YarnException"/>
protected internal override void SubmitApplication(ApplicationSubmissionContext submissionContext
, long submitTime, string user)
{
//Do nothing, just add the application to RMContext
RMAppImpl application = new RMAppImpl(submissionContext.GetApplicationId(), this.
rmContext, this.conf, submissionContext.GetApplicationName(), user, submissionContext
.GetQueue(), submissionContext, this.rmContext.GetScheduler(), this.rmContext.GetApplicationMasterService
(), submitTime, submissionContext.GetApplicationType(), submissionContext.GetApplicationTags
(), null);
this.rmContext.GetRMApps()[submissionContext.GetApplicationId()] = application;
}
/// <exception cref="System.Exception"/>
protected internal virtual void RecoverApplication(ApplicationStateData appState,
RMStateStore.RMState rmState)
{
ApplicationSubmissionContext appContext = appState.GetApplicationSubmissionContext
();
ApplicationId appId = appContext.GetApplicationId();
// create and recover app.
RMAppImpl application = CreateAndPopulateNewRMApp(appContext, appState.GetSubmitTime
(), appState.GetUser(), true);
application.Handle(new RMAppRecoverEvent(appId, rmState));
}
protected internal virtual void CreateApplicationWithAMResource(ApplicationAttemptId
attId, string queue, string user, Resource amResource)
{
RMContext rmContext = resourceManager.GetRMContext();
RMApp rmApp = new RMAppImpl(attId.GetApplicationId(), rmContext, conf, null, null
, null, ApplicationSubmissionContext.NewInstance(null, null, null, null, null, false
, false, 0, amResource, null), null, null, 0, null, null, null);
rmContext.GetRMApps()[attId.GetApplicationId()] = rmApp;
AppAddedSchedulerEvent appAddedEvent = new AppAddedSchedulerEvent(attId.GetApplicationId
(), queue, user);
scheduler.Handle(appAddedEvent);
AppAttemptAddedSchedulerEvent attempAddedEvent = new AppAttemptAddedSchedulerEvent
(attId, false);
scheduler.Handle(attempAddedEvent);
}
private RMAppImpl CreateMockRMApp(ApplicationAttemptId attemptId, RMContext context
)
{
RMAppImpl app = Org.Mockito.Mockito.Mock <RMAppImpl>();
Org.Mockito.Mockito.When(app.GetApplicationId()).ThenReturn(attemptId.GetApplicationId
());
RMAppAttemptImpl attempt = Org.Mockito.Mockito.Mock <RMAppAttemptImpl>();
Org.Mockito.Mockito.When(attempt.GetAppAttemptId()).ThenReturn(attemptId);
RMAppAttemptMetrics attemptMetric = Org.Mockito.Mockito.Mock <RMAppAttemptMetrics>
();
Org.Mockito.Mockito.When(attempt.GetRMAppAttemptMetrics()).ThenReturn(attemptMetric
);
Org.Mockito.Mockito.When(app.GetCurrentAppAttempt()).ThenReturn(attempt);
context.GetRMApps().PutIfAbsent(attemptId.GetApplicationId(), app);
return(app);
}
/// <exception cref="Org.Apache.Hadoop.Yarn.Exceptions.YarnException"/>
protected internal virtual void SubmitApplication(ApplicationSubmissionContext submissionContext
, long submitTime, string user)
{
ApplicationId applicationId = submissionContext.GetApplicationId();
RMAppImpl application = CreateAndPopulateNewRMApp(submissionContext, submitTime,
user, false);
ApplicationId appId = submissionContext.GetApplicationId();
if (UserGroupInformation.IsSecurityEnabled())
{
try
{
this.rmContext.GetDelegationTokenRenewer().AddApplicationAsync(appId, ParseCredentials
(submissionContext), submissionContext.GetCancelTokensWhenComplete(), application
.GetUser());
}
catch (Exception e)
{
Log.Warn("Unable to parse credentials.", e);
// Sending APP_REJECTED is fine, since we assume that the
// RMApp is in NEW state and thus we haven't yet informed the
// scheduler about the existence of the application
System.Diagnostics.Debug.Assert(application.GetState() == RMAppState.New);
this.rmContext.GetDispatcher().GetEventHandler().Handle(new RMAppEvent(applicationId
, RMAppEventType.AppRejected, e.Message));
throw RPCUtil.GetRemoteException(e);
}
}
else
{
// Dispatcher is not yet started at this time, so these START events
// enqueued should be guaranteed to be first processed when dispatcher
// gets started.
this.rmContext.GetDispatcher().GetEventHandler().Handle(new RMAppEvent(applicationId
, RMAppEventType.Start));
}
}
/// <exception cref="System.Exception"/>
public virtual void TestRMAppAttemptFailuresValidityInterval()
{
YarnConfiguration conf = new YarnConfiguration();
conf.SetClass(YarnConfiguration.RmScheduler, typeof(CapacityScheduler), typeof(ResourceScheduler
));
conf.SetBoolean(YarnConfiguration.RecoveryEnabled, true);
conf.SetBoolean(YarnConfiguration.RmWorkPreservingRecoveryEnabled, false);
conf.Set(YarnConfiguration.RmStore, typeof(MemoryRMStateStore).FullName);
// explicitly set max-am-retry count as 2.
conf.SetInt(YarnConfiguration.RmAmMaxAttempts, 2);
MemoryRMStateStore memStore = new MemoryRMStateStore();
memStore.Init(conf);
MockRM rm1 = new MockRM(conf, memStore);
rm1.Start();
MockNM nm1 = new MockNM("127.0.0.1:1234", 8000, rm1.GetResourceTrackerService());
nm1.RegisterNode();
// set window size to a larger number : 20s
// we will verify the app should be failed if
// two continuous attempts failed in 20s.
RMApp app = rm1.SubmitApp(200, 20000);
MockAM am = MockRM.LaunchAM(app, rm1, nm1);
// Fail current attempt normally
nm1.NodeHeartbeat(am.GetApplicationAttemptId(), 1, ContainerState.Complete);
am.WaitForState(RMAppAttemptState.Failed);
// launch the second attempt
rm1.WaitForState(app.GetApplicationId(), RMAppState.Accepted);
NUnit.Framework.Assert.AreEqual(2, app.GetAppAttempts().Count);
NUnit.Framework.Assert.IsTrue(((RMAppAttemptImpl)app.GetCurrentAppAttempt()).MayBeLastAttempt
());
MockAM am_2 = MockRM.LaunchAndRegisterAM(app, rm1, nm1);
am_2.WaitForState(RMAppAttemptState.Running);
nm1.NodeHeartbeat(am_2.GetApplicationAttemptId(), 1, ContainerState.Complete);
am_2.WaitForState(RMAppAttemptState.Failed);
// current app should be failed.
rm1.WaitForState(app.GetApplicationId(), RMAppState.Failed);
ControlledClock clock = new ControlledClock(new SystemClock());
// set window size to 6s
RMAppImpl app1 = (RMAppImpl)rm1.SubmitApp(200, 6000);
app1.SetSystemClock(clock);
MockAM am1 = MockRM.LaunchAndRegisterAM(app1, rm1, nm1);
// Fail attempt1 normally
nm1.NodeHeartbeat(am1.GetApplicationAttemptId(), 1, ContainerState.Complete);
am1.WaitForState(RMAppAttemptState.Failed);
// launch the second attempt
rm1.WaitForState(app1.GetApplicationId(), RMAppState.Accepted);
NUnit.Framework.Assert.AreEqual(2, app1.GetAppAttempts().Count);
RMAppAttempt attempt2 = app1.GetCurrentAppAttempt();
NUnit.Framework.Assert.IsTrue(((RMAppAttemptImpl)attempt2).MayBeLastAttempt());
MockAM am2 = MockRM.LaunchAndRegisterAM(app1, rm1, nm1);
am2.WaitForState(RMAppAttemptState.Running);
// wait for 6 seconds
clock.SetTime(Runtime.CurrentTimeMillis() + 6 * 1000);
// Fail attempt2 normally
nm1.NodeHeartbeat(am2.GetApplicationAttemptId(), 1, ContainerState.Complete);
am2.WaitForState(RMAppAttemptState.Failed);
// can launch the third attempt successfully
rm1.WaitForState(app1.GetApplicationId(), RMAppState.Accepted);
NUnit.Framework.Assert.AreEqual(3, app1.GetAppAttempts().Count);
RMAppAttempt attempt3 = app1.GetCurrentAppAttempt();
clock.Reset();
MockAM am3 = MockRM.LaunchAndRegisterAM(app1, rm1, nm1);
am3.WaitForState(RMAppAttemptState.Running);
// Restart rm.
MockRM rm2 = new MockRM(conf, memStore);
rm2.Start();
// re-register the NM
nm1.SetResourceTrackerService(rm2.GetResourceTrackerService());
NMContainerStatus status = Org.Apache.Hadoop.Yarn.Util.Records.NewRecord <NMContainerStatus
>();
status.SetContainerExitStatus(ContainerExitStatus.KilledByResourcemanager);
status.SetContainerId(attempt3.GetMasterContainer().GetId());
status.SetContainerState(ContainerState.Complete);
status.SetDiagnostics(string.Empty);
nm1.RegisterNode(Sharpen.Collections.SingletonList(status), null);
rm2.WaitForState(attempt3.GetAppAttemptId(), RMAppAttemptState.Failed);
rm2.WaitForState(app1.GetApplicationId(), RMAppState.Accepted);
// Lauch Attempt 4
MockAM am4 = rm2.WaitForNewAMToLaunchAndRegister(app1.GetApplicationId(), 4, nm1);
// wait for 6 seconds
clock.SetTime(Runtime.CurrentTimeMillis() + 6 * 1000);
// Fail attempt4 normally
nm1.NodeHeartbeat(am4.GetApplicationAttemptId(), 1, ContainerState.Complete);
am4.WaitForState(RMAppAttemptState.Failed);
// can launch the 5th attempt successfully
rm2.WaitForState(app1.GetApplicationId(), RMAppState.Accepted);
MockAM am5 = rm2.WaitForNewAMToLaunchAndRegister(app1.GetApplicationId(), 5, nm1);
clock.Reset();
am5.WaitForState(RMAppAttemptState.Running);
// Fail attempt5 normally
nm1.NodeHeartbeat(am5.GetApplicationAttemptId(), 1, ContainerState.Complete);
am5.WaitForState(RMAppAttemptState.Failed);
rm2.WaitForState(app1.GetApplicationId(), RMAppState.Failed);
rm1.Stop();
rm2.Stop();
}
