public virtual void TestAllcoateRequestWithIncrease()
{
IList <ContainerResourceIncreaseRequest> incRequests = new AList <ContainerResourceIncreaseRequest
>();
for (int i = 0; i < 3; i++)
{
incRequests.AddItem(ContainerResourceIncreaseRequest.NewInstance(null, Resource.NewInstance
(0, i)));
}
AllocateRequest r = AllocateRequest.NewInstance(123, 0f, null, null, null, incRequests
);
// serde
YarnServiceProtos.AllocateRequestProto p = ((AllocateRequestPBImpl)r).GetProto();
r = new AllocateRequestPBImpl(p);
// check value
NUnit.Framework.Assert.AreEqual(123, r.GetResponseId());
NUnit.Framework.Assert.AreEqual(incRequests.Count, r.GetIncreaseRequests().Count);
for (int i_1 = 0; i_1 < incRequests.Count; i_1++)
{
NUnit.Framework.Assert.AreEqual(r.GetIncreaseRequests()[i_1].GetCapability().GetVirtualCores
(), incRequests[i_1].GetCapability().GetVirtualCores());
}
}
/// <exception cref="System.Exception"/>
public virtual AllocateResponse Allocate(IList <ResourceRequest> resourceRequest,
IList <ContainerId> releases)
{
AllocateRequest req = AllocateRequest.NewInstance(0, 0F, resourceRequest, releases
, null);
return(Allocate(req));
}
/// <exception cref="System.Exception"/>
protected internal override void Heartbeat()
{
lock (this)
{
AllocateRequest allocateRequest = AllocateRequest.NewInstance(this.lastResponseID
, base.GetApplicationProgress(), new AList <ResourceRequest>(), new AList <ContainerId
>(), null);
AllocateResponse allocateResponse = null;
try
{
allocateResponse = scheduler.Allocate(allocateRequest);
// Reset retry count if no exception occurred.
retrystartTime = Runtime.CurrentTimeMillis();
}
catch (ApplicationAttemptNotFoundException e)
{
Log.Info("Event from RM: shutting down Application Master");
// This can happen if the RM has been restarted. If it is in that state,
// this application must clean itself up.
eventHandler.Handle(new JobEvent(this.GetJob().GetID(), JobEventType.JobAmReboot)
);
throw new YarnRuntimeException("Resource Manager doesn't recognize AttemptId: " +
this.GetContext().GetApplicationID(), e);
}
catch (ApplicationMasterNotRegisteredException)
{
Log.Info("ApplicationMaster is out of sync with ResourceManager," + " hence resync and send outstanding requests."
);
this.lastResponseID = 0;
Register();
}
catch (Exception e)
{
// This can happen when the connection to the RM has gone down. Keep
// re-trying until the retryInterval has expired.
if (Runtime.CurrentTimeMillis() - retrystartTime >= retryInterval)
{
Log.Error("Could not contact RM after " + retryInterval + " milliseconds.");
eventHandler.Handle(new JobEvent(this.GetJob().GetID(), JobEventType.InternalError
));
throw new YarnRuntimeException("Could not contact RM after " + retryInterval + " milliseconds."
);
}
// Throw this up to the caller, which may decide to ignore it and
// continue to attempt to contact the RM.
throw;
}
if (allocateResponse != null)
{
this.lastResponseID = allocateResponse.GetResponseId();
Token token = allocateResponse.GetAMRMToken();
if (token != null)
{
UpdateAMRMToken(token);
}
}
}
}
/// <exception cref="Org.Apache.Hadoop.Yarn.Exceptions.YarnException"/>
/// <exception cref="System.IO.IOException"/>
public virtual void TestAllocateOnHA()
{
AllocateRequest request = AllocateRequest.NewInstance(0, 50f, new AList <ResourceRequest
>(), new AList <ContainerId>(), ResourceBlacklistRequest.NewInstance(new AList <string
>(), new AList <string>()));
AllocateResponse response = amClient.Allocate(request);
NUnit.Framework.Assert.AreEqual(response, this.cluster.CreateFakeAllocateResponse
());
}
public virtual void TestAllcoateRequestWithoutIncrease()
{
AllocateRequest r = AllocateRequest.NewInstance(123, 0f, null, null, null, null);
// serde
YarnServiceProtos.AllocateRequestProto p = ((AllocateRequestPBImpl)r).GetProto();
r = new AllocateRequestPBImpl(p);
// check value
NUnit.Framework.Assert.AreEqual(123, r.GetResponseId());
NUnit.Framework.Assert.AreEqual(0, r.GetIncreaseRequests().Count);
}
public virtual void TestValidateResourceBlacklistRequest()
{
TestAMAuthorization.MyContainerManager containerManager = new TestAMAuthorization.MyContainerManager
();
TestAMAuthorization.MockRMWithAMS rm = new TestAMAuthorization.MockRMWithAMS(new
YarnConfiguration(), containerManager);
rm.Start();
MockNM nm1 = rm.RegisterNode("localhost:1234", 5120);
IDictionary <ApplicationAccessType, string> acls = new Dictionary <ApplicationAccessType
, string>(2);
acls[ApplicationAccessType.ViewApp] = "*";
RMApp app = rm.SubmitApp(1024, "appname", "appuser", acls);
nm1.NodeHeartbeat(true);
RMAppAttempt attempt = app.GetCurrentAppAttempt();
ApplicationAttemptId applicationAttemptId = attempt.GetAppAttemptId();
WaitForLaunchedState(attempt);
// Create a client to the RM.
Configuration conf = rm.GetConfig();
YarnRPC rpc = YarnRPC.Create(conf);
UserGroupInformation currentUser = UserGroupInformation.CreateRemoteUser(applicationAttemptId
.ToString());
Credentials credentials = containerManager.GetContainerCredentials();
IPEndPoint rmBindAddress = rm.GetApplicationMasterService().GetBindAddress();
Org.Apache.Hadoop.Security.Token.Token <TokenIdentifier> amRMToken = TestAMAuthorization.MockRMWithAMS
.SetupAndReturnAMRMToken(rmBindAddress, credentials.GetAllTokens());
currentUser.AddToken(amRMToken);
ApplicationMasterProtocol client = currentUser.DoAs(new _PrivilegedAction_626(rpc
, rmBindAddress, conf));
RegisterApplicationMasterRequest request = Org.Apache.Hadoop.Yarn.Util.Records.NewRecord
<RegisterApplicationMasterRequest>();
client.RegisterApplicationMaster(request);
ResourceBlacklistRequest blacklistRequest = ResourceBlacklistRequest.NewInstance(
Sharpen.Collections.SingletonList(ResourceRequest.Any), null);
AllocateRequest allocateRequest = AllocateRequest.NewInstance(0, 0.0f, null, null
, blacklistRequest);
bool error = false;
try
{
client.Allocate(allocateRequest);
}
catch (InvalidResourceBlacklistRequestException)
{
error = true;
}
rm.Stop();
NUnit.Framework.Assert.IsTrue("Didn't not catch InvalidResourceBlacklistRequestException"
, error);
}
/// <exception cref="Org.Apache.Hadoop.Yarn.Exceptions.YarnException"/>
/// <exception cref="System.IO.IOException"/>
protected internal virtual AllocateResponse MakeRemoteRequest()
{
ApplyRequestLimits();
ResourceBlacklistRequest blacklistRequest = ResourceBlacklistRequest.NewInstance(
new AList <string>(blacklistAdditions), new AList <string>(blacklistRemovals));
AllocateRequest allocateRequest = AllocateRequest.NewInstance(lastResponseID, base
.GetApplicationProgress(), new AList <ResourceRequest>(ask), new AList <ContainerId
>(release), blacklistRequest);
AllocateResponse allocateResponse = scheduler.Allocate(allocateRequest);
lastResponseID = allocateResponse.GetResponseId();
availableResources = allocateResponse.GetAvailableResources();
lastClusterNmCount = clusterNmCount;
clusterNmCount = allocateResponse.GetNumClusterNodes();
int numCompletedContainers = allocateResponse.GetCompletedContainersStatuses().Count;
if (ask.Count > 0 || release.Count > 0)
{
Log.Info("getResources() for " + applicationId + ":" + " ask=" + ask.Count + " release= "
+ release.Count + " newContainers=" + allocateResponse.GetAllocatedContainers()
.Count + " finishedContainers=" + numCompletedContainers + " resourcelimit=" + availableResources
+ " knownNMs=" + clusterNmCount);
}
ask.Clear();
release.Clear();
if (numCompletedContainers > 0)
{
// re-send limited requests when a container completes to trigger asking
// for more containers
Sharpen.Collections.AddAll(requestLimitsToUpdate, requestLimits.Keys);
}
if (blacklistAdditions.Count > 0 || blacklistRemovals.Count > 0)
{
Log.Info("Update the blacklist for " + applicationId + ": blacklistAdditions=" +
blacklistAdditions.Count + " blacklistRemovals=" + blacklistRemovals.Count);
}
blacklistAdditions.Clear();
blacklistRemovals.Clear();
return(allocateResponse);
}
public virtual void TestARRMResponseId()
{
MockNM nm1 = rm.RegisterNode("h1:1234", 5000);
RMApp app = rm.SubmitApp(2000);
// Trigger the scheduling so the AM gets 'launched'
nm1.NodeHeartbeat(true);
RMAppAttempt attempt = app.GetCurrentAppAttempt();
MockAM am = rm.SendAMLaunched(attempt.GetAppAttemptId());
am.RegisterAppAttempt();
AllocateRequest allocateRequest = AllocateRequest.NewInstance(0, 0F, null, null,
null);
AllocateResponse response = Allocate(attempt.GetAppAttemptId(), allocateRequest);
NUnit.Framework.Assert.AreEqual(1, response.GetResponseId());
NUnit.Framework.Assert.IsTrue(response.GetAMCommand() == null);
allocateRequest = AllocateRequest.NewInstance(response.GetResponseId(), 0F, null,
null, null);
response = Allocate(attempt.GetAppAttemptId(), allocateRequest);
NUnit.Framework.Assert.AreEqual(2, response.GetResponseId());
/* try resending */
response = Allocate(attempt.GetAppAttemptId(), allocateRequest);
NUnit.Framework.Assert.AreEqual(2, response.GetResponseId());
allocateRequest = AllocateRequest.NewInstance(0, 0F, null, null, null);
try
{
Allocate(attempt.GetAppAttemptId(), allocateRequest);
NUnit.Framework.Assert.Fail();
}
catch (Exception e)
{
NUnit.Framework.Assert.IsTrue(e.InnerException is InvalidApplicationMasterRequestException
);
}
}
// The test verifies processing of NMContainerStatuses which are sent during
// NM registration.
// 1. Start the cluster-RM,NM,Submit app with 1024MB,Launch & register AM
// 2. AM sends ResourceRequest for 1 container with memory 2048MB.
// 3. Verify for number of container allocated by RM
// 4. Verify Memory Usage by cluster, it should be 3072. AM memory + requested
// memory. 1024 + 2048=3072
// 5. Re-register NM by sending completed container status
// 6. Verify for Memory Used, it should be 1024
// 7. Send AM heatbeat to RM. Allocated response should contain completed
// container.
/// <exception cref="System.Exception"/>
public virtual void TestProcessingNMContainerStatusesOnNMRestart()
{
conf.SetInt(YarnConfiguration.RmAmMaxAttempts, 1);
MemoryRMStateStore memStore = new MemoryRMStateStore();
memStore.Init(conf);
// 1. Start the cluster-RM,NM,Submit app with 1024MB,Launch & register AM
MockRM rm1 = new MockRM(conf, memStore);
rm1.Start();
int nmMemory = 8192;
int amMemory = 1024;
int containerMemory = 2048;
MockNM nm1 = new MockNM("127.0.0.1:1234", nmMemory, rm1.GetResourceTrackerService
());
nm1.RegisterNode();
RMApp app0 = rm1.SubmitApp(amMemory);
MockAM am0 = MockRM.LaunchAndRegisterAM(app0, rm1, nm1);
// 2. AM sends ResourceRequest for 1 container with memory 2048MB.
int noOfContainers = 1;
IList <Container> allocateContainers = am0.AllocateAndWaitForContainers(noOfContainers
, containerMemory, nm1);
// 3. Verify for number of container allocated by RM
NUnit.Framework.Assert.AreEqual(noOfContainers, allocateContainers.Count);
Container container = allocateContainers[0];
nm1.NodeHeartbeat(am0.GetApplicationAttemptId(), 1, ContainerState.Running);
nm1.NodeHeartbeat(am0.GetApplicationAttemptId(), container.GetId().GetContainerId
(), ContainerState.Running);
rm1.WaitForState(app0.GetApplicationId(), RMAppState.Running);
// 4. Verify Memory Usage by cluster, it should be 3072. AM memory +
// requested memory. 1024 + 2048=3072
ResourceScheduler rs = rm1.GetRMContext().GetScheduler();
int allocatedMB = rs.GetRootQueueMetrics().GetAllocatedMB();
NUnit.Framework.Assert.AreEqual(amMemory + containerMemory, allocatedMB);
// 5. Re-register NM by sending completed container status
IList <NMContainerStatus> nMContainerStatusForApp = CreateNMContainerStatusForApp(
am0);
nm1.RegisterNode(nMContainerStatusForApp, Arrays.AsList(app0.GetApplicationId()));
WaitForClusterMemory(nm1, rs, amMemory);
// 6. Verify for Memory Used, it should be 1024
NUnit.Framework.Assert.AreEqual(amMemory, rs.GetRootQueueMetrics().GetAllocatedMB
());
// 7. Send AM heatbeat to RM. Allocated response should contain completed
// container
AllocateRequest req = AllocateRequest.NewInstance(0, 0F, new AList <ResourceRequest
>(), new AList <ContainerId>(), null);
AllocateResponse allocate = am0.Allocate(req);
IList <ContainerStatus> completedContainersStatuses = allocate.GetCompletedContainersStatuses
();
NUnit.Framework.Assert.AreEqual(noOfContainers, completedContainersStatuses.Count
);
// Application clean up should happen Cluster memory used is 0
nm1.NodeHeartbeat(am0.GetApplicationAttemptId(), 1, ContainerState.Complete);
WaitForClusterMemory(nm1, rs, 0);
rm1.Stop();
}
/// <exception cref="Org.Apache.Hadoop.Yarn.Exceptions.YarnException"/>
/// <exception cref="System.IO.IOException"/>
public override AllocateResponse Allocate(float progressIndicator)
{
Preconditions.CheckArgument(progressIndicator >= 0, "Progress indicator should not be negative"
);
AllocateResponse allocateResponse = null;
IList <ResourceRequest> askList = null;
IList <ContainerId> releaseList = null;
AllocateRequest allocateRequest = null;
IList <string> blacklistToAdd = new AList <string>();
IList <string> blacklistToRemove = new AList <string>();
try
{
lock (this)
{
askList = new AList <ResourceRequest>(ask.Count);
foreach (ResourceRequest r in ask)
{
// create a copy of ResourceRequest as we might change it while the
// RPC layer is using it to send info across
askList.AddItem(ResourceRequest.NewInstance(r.GetPriority(), r.GetResourceName(),
r.GetCapability(), r.GetNumContainers(), r.GetRelaxLocality(), r.GetNodeLabelExpression
()));
}
releaseList = new AList <ContainerId>(release);
// optimistically clear this collection assuming no RPC failure
ask.Clear();
release.Clear();
Sharpen.Collections.AddAll(blacklistToAdd, blacklistAdditions);
Sharpen.Collections.AddAll(blacklistToRemove, blacklistRemovals);
ResourceBlacklistRequest blacklistRequest = ResourceBlacklistRequest.NewInstance(
blacklistToAdd, blacklistToRemove);
allocateRequest = AllocateRequest.NewInstance(lastResponseId, progressIndicator,
askList, releaseList, blacklistRequest);
// clear blacklistAdditions and blacklistRemovals before
// unsynchronized part
blacklistAdditions.Clear();
blacklistRemovals.Clear();
}
try
{
allocateResponse = rmClient.Allocate(allocateRequest);
}
catch (ApplicationMasterNotRegisteredException)
{
Log.Warn("ApplicationMaster is out of sync with ResourceManager," + " hence resyncing."
);
lock (this)
{
Sharpen.Collections.AddAll(release, this.pendingRelease);
Sharpen.Collections.AddAll(blacklistAdditions, this.blacklistedNodes);
foreach (IDictionary <string, SortedDictionary <Resource, AMRMClientImpl.ResourceRequestInfo
> > rr in remoteRequestsTable.Values)
{
foreach (IDictionary <Resource, AMRMClientImpl.ResourceRequestInfo> capabalities in
rr.Values)
{
foreach (AMRMClientImpl.ResourceRequestInfo request in capabalities.Values)
{
AddResourceRequestToAsk(request.remoteRequest);
}
}
}
}
// re register with RM
RegisterApplicationMaster();
allocateResponse = Allocate(progressIndicator);
return(allocateResponse);
}
lock (this)
{
// update these on successful RPC
clusterNodeCount = allocateResponse.GetNumClusterNodes();
lastResponseId = allocateResponse.GetResponseId();
clusterAvailableResources = allocateResponse.GetAvailableResources();
if (!allocateResponse.GetNMTokens().IsEmpty())
{
PopulateNMTokens(allocateResponse.GetNMTokens());
}
if (allocateResponse.GetAMRMToken() != null)
{
UpdateAMRMToken(allocateResponse.GetAMRMToken());
}
if (!pendingRelease.IsEmpty() && !allocateResponse.GetCompletedContainersStatuses
().IsEmpty())
{
RemovePendingReleaseRequests(allocateResponse.GetCompletedContainersStatuses());
}
}
}
finally
{
// TODO how to differentiate remote yarn exception vs error in rpc
if (allocateResponse == null)
{
// we hit an exception in allocate()
// preserve ask and release for next call to allocate()
lock (this)
{
Sharpen.Collections.AddAll(release, releaseList);
// requests could have been added or deleted during call to allocate
// If requests were added/removed then there is nothing to do since
// the ResourceRequest object in ask would have the actual new value.
// If ask does not have this ResourceRequest then it was unchanged and
// so we can add the value back safely.
// This assumes that there will no concurrent calls to allocate() and
// so we dont have to worry about ask being changed in the
// synchronized block at the beginning of this method.
foreach (ResourceRequest oldAsk in askList)
{
if (!ask.Contains(oldAsk))
{
ask.AddItem(oldAsk);
}
}
Sharpen.Collections.AddAll(blacklistAdditions, blacklistToAdd);
Sharpen.Collections.AddAll(blacklistRemovals, blacklistToRemove);
}
}
}
return(allocateResponse);
}
public virtual void TestAMRMUnusableNodes()
{
MockNM nm1 = rm.RegisterNode("127.0.0.1:1234", 10000);
MockNM nm2 = rm.RegisterNode("127.0.0.2:1234", 10000);
MockNM nm3 = rm.RegisterNode("127.0.0.3:1234", 10000);
MockNM nm4 = rm.RegisterNode("127.0.0.4:1234", 10000);
dispatcher.Await();
RMApp app1 = rm.SubmitApp(2000);
// Trigger the scheduling so the AM gets 'launched' on nm1
nm1.NodeHeartbeat(true);
RMAppAttempt attempt1 = app1.GetCurrentAppAttempt();
MockAM am1 = rm.SendAMLaunched(attempt1.GetAppAttemptId());
// register AM returns no unusable node
am1.RegisterAppAttempt();
// allocate request returns no updated node
AllocateRequest allocateRequest1 = AllocateRequest.NewInstance(0, 0F, null, null,
null);
AllocateResponse response1 = Allocate(attempt1.GetAppAttemptId(), allocateRequest1
);
IList <NodeReport> updatedNodes = response1.GetUpdatedNodes();
NUnit.Framework.Assert.AreEqual(0, updatedNodes.Count);
SyncNodeHeartbeat(nm4, false);
// allocate request returns updated node
allocateRequest1 = AllocateRequest.NewInstance(response1.GetResponseId(), 0F, null
, null, null);
response1 = Allocate(attempt1.GetAppAttemptId(), allocateRequest1);
updatedNodes = response1.GetUpdatedNodes();
NUnit.Framework.Assert.AreEqual(1, updatedNodes.Count);
NodeReport nr = updatedNodes.GetEnumerator().Next();
NUnit.Framework.Assert.AreEqual(nm4.GetNodeId(), nr.GetNodeId());
NUnit.Framework.Assert.AreEqual(NodeState.Unhealthy, nr.GetNodeState());
// resending the allocate request returns the same result
response1 = Allocate(attempt1.GetAppAttemptId(), allocateRequest1);
updatedNodes = response1.GetUpdatedNodes();
NUnit.Framework.Assert.AreEqual(1, updatedNodes.Count);
nr = updatedNodes.GetEnumerator().Next();
NUnit.Framework.Assert.AreEqual(nm4.GetNodeId(), nr.GetNodeId());
NUnit.Framework.Assert.AreEqual(NodeState.Unhealthy, nr.GetNodeState());
SyncNodeLost(nm3);
// subsequent allocate request returns delta
allocateRequest1 = AllocateRequest.NewInstance(response1.GetResponseId(), 0F, null
, null, null);
response1 = Allocate(attempt1.GetAppAttemptId(), allocateRequest1);
updatedNodes = response1.GetUpdatedNodes();
NUnit.Framework.Assert.AreEqual(1, updatedNodes.Count);
nr = updatedNodes.GetEnumerator().Next();
NUnit.Framework.Assert.AreEqual(nm3.GetNodeId(), nr.GetNodeId());
NUnit.Framework.Assert.AreEqual(NodeState.Lost, nr.GetNodeState());
// registering another AM gives it the complete failed list
RMApp app2 = rm.SubmitApp(2000);
// Trigger nm2 heartbeat so that AM gets launched on it
nm2.NodeHeartbeat(true);
RMAppAttempt attempt2 = app2.GetCurrentAppAttempt();
MockAM am2 = rm.SendAMLaunched(attempt2.GetAppAttemptId());
// register AM returns all unusable nodes
am2.RegisterAppAttempt();
// allocate request returns no updated node
AllocateRequest allocateRequest2 = AllocateRequest.NewInstance(0, 0F, null, null,
null);
AllocateResponse response2 = Allocate(attempt2.GetAppAttemptId(), allocateRequest2
);
updatedNodes = response2.GetUpdatedNodes();
NUnit.Framework.Assert.AreEqual(0, updatedNodes.Count);
SyncNodeHeartbeat(nm4, true);
// both AM's should get delta updated nodes
allocateRequest1 = AllocateRequest.NewInstance(response1.GetResponseId(), 0F, null
, null, null);
response1 = Allocate(attempt1.GetAppAttemptId(), allocateRequest1);
updatedNodes = response1.GetUpdatedNodes();
NUnit.Framework.Assert.AreEqual(1, updatedNodes.Count);
nr = updatedNodes.GetEnumerator().Next();
NUnit.Framework.Assert.AreEqual(nm4.GetNodeId(), nr.GetNodeId());
NUnit.Framework.Assert.AreEqual(NodeState.Running, nr.GetNodeState());
allocateRequest2 = AllocateRequest.NewInstance(response2.GetResponseId(), 0F, null
, null, null);
response2 = Allocate(attempt2.GetAppAttemptId(), allocateRequest2);
updatedNodes = response2.GetUpdatedNodes();
NUnit.Framework.Assert.AreEqual(1, updatedNodes.Count);
nr = updatedNodes.GetEnumerator().Next();
NUnit.Framework.Assert.AreEqual(nm4.GetNodeId(), nr.GetNodeId());
NUnit.Framework.Assert.AreEqual(NodeState.Running, nr.GetNodeState());
// subsequent allocate calls should return no updated nodes
allocateRequest2 = AllocateRequest.NewInstance(response2.GetResponseId(), 0F, null
, null, null);
response2 = Allocate(attempt2.GetAppAttemptId(), allocateRequest2);
updatedNodes = response2.GetUpdatedNodes();
NUnit.Framework.Assert.AreEqual(0, updatedNodes.Count);
}