/// <summary>Implements getLiveNodeManagers()</summary>
/// <returns>JSON formatted string containing statuses of all node managers</returns>
public virtual string GetLiveNodeManagers()
{
// RMNMInfoBeans
ICollection <RMNode> nodes = this.rmContext.GetRMNodes().Values;
IList <RMNMInfo.InfoMap> nodesInfo = new AList <RMNMInfo.InfoMap>();
foreach (RMNode ni in nodes)
{
SchedulerNodeReport report = scheduler.GetNodeReport(ni.GetNodeID());
RMNMInfo.InfoMap info = new RMNMInfo.InfoMap();
info["HostName"] = ni.GetHostName();
info["Rack"] = ni.GetRackName();
info["State"] = ni.GetState().ToString();
info["NodeId"] = ni.GetNodeID();
info["NodeHTTPAddress"] = ni.GetHttpAddress();
info["LastHealthUpdate"] = ni.GetLastHealthReportTime();
info["HealthReport"] = ni.GetHealthReport();
info["NodeManagerVersion"] = ni.GetNodeManagerVersion();
if (report != null)
{
info["NumContainers"] = report.GetNumContainers();
info["UsedMemoryMB"] = report.GetUsedResource().GetMemory();
info["AvailableMemoryMB"] = report.GetAvailableResource().GetMemory();
}
nodesInfo.AddItem(info);
}
return(JSON.ToString(nodesInfo));
}
public NodeInfo(RMNode ni, ResourceScheduler sched)
{
// JAXB needs this
NodeId id = ni.GetNodeID();
SchedulerNodeReport report = sched.GetNodeReport(id);
this.numContainers = 0;
this.usedMemoryMB = 0;
this.availMemoryMB = 0;
if (report != null)
{
this.numContainers = report.GetNumContainers();
this.usedMemoryMB = report.GetUsedResource().GetMemory();
this.availMemoryMB = report.GetAvailableResource().GetMemory();
this.usedVirtualCores = report.GetUsedResource().GetVirtualCores();
this.availableVirtualCores = report.GetAvailableResource().GetVirtualCores();
}
this.id = id.ToString();
this.rack = ni.GetRackName();
this.nodeHostName = ni.GetHostName();
this.state = ni.GetState();
this.nodeHTTPAddress = ni.GetHttpAddress();
this.lastHealthUpdate = ni.GetLastHealthReportTime();
this.healthReport = ni.GetHealthReport().ToString();
this.version = ni.GetNodeManagerVersion();
// add labels
ICollection <string> labelSet = ni.GetNodeLabels();
if (labelSet != null)
{
Sharpen.Collections.AddAll(nodeLabels, labelSet);
nodeLabels.Sort();
}
}
/// <exception cref="Org.Codehaus.Jettison.Json.JSONException"/>
/// <exception cref="System.Exception"/>
public virtual void VerifyNodeInfoGeneric(MockNM nm, string state, string rack, string
id, string nodeHostName, string nodeHTTPAddress, long lastHealthUpdate, string
healthReport, int numContainers, long usedMemoryMB, long availMemoryMB, long usedVirtualCores
, long availVirtualCores, string version)
{
RMNode node = rm.GetRMContext().GetRMNodes()[nm.GetNodeId()];
ResourceScheduler sched = rm.GetResourceScheduler();
SchedulerNodeReport report = sched.GetNodeReport(nm.GetNodeId());
WebServicesTestUtils.CheckStringMatch("state", node.GetState().ToString(), state);
WebServicesTestUtils.CheckStringMatch("rack", node.GetRackName(), rack);
WebServicesTestUtils.CheckStringMatch("id", nm.GetNodeId().ToString(), id);
WebServicesTestUtils.CheckStringMatch("nodeHostName", nm.GetNodeId().GetHost(), nodeHostName
);
WebServicesTestUtils.CheckStringMatch("healthReport", node.GetHealthReport().ToString
(), healthReport);
string expectedHttpAddress = nm.GetNodeId().GetHost() + ":" + nm.GetHttpPort();
WebServicesTestUtils.CheckStringMatch("nodeHTTPAddress", expectedHttpAddress, nodeHTTPAddress
);
WebServicesTestUtils.CheckStringMatch("version", node.GetNodeManagerVersion(), version
);
long expectedHealthUpdate = node.GetLastHealthReportTime();
NUnit.Framework.Assert.AreEqual("lastHealthUpdate doesn't match, got: " + lastHealthUpdate
+ " expected: " + expectedHealthUpdate, expectedHealthUpdate, lastHealthUpdate);
if (report != null)
{
NUnit.Framework.Assert.AreEqual("numContainers doesn't match: " + numContainers,
report.GetNumContainers(), numContainers);
NUnit.Framework.Assert.AreEqual("usedMemoryMB doesn't match: " + usedMemoryMB, report
.GetUsedResource().GetMemory(), usedMemoryMB);
NUnit.Framework.Assert.AreEqual("availMemoryMB doesn't match: " + availMemoryMB,
report.GetAvailableResource().GetMemory(), availMemoryMB);
NUnit.Framework.Assert.AreEqual("usedVirtualCores doesn't match: " + usedVirtualCores
, report.GetUsedResource().GetVirtualCores(), usedVirtualCores);
NUnit.Framework.Assert.AreEqual("availVirtualCores doesn't match: " + availVirtualCores
, report.GetAvailableResource().GetVirtualCores(), availVirtualCores);
}
}
/// <exception cref="Org.Apache.Hadoop.Yarn.Exceptions.YarnException"/>
/// <exception cref="System.IO.IOException"/>
public virtual AllocateResponse Allocate(AllocateRequest request)
{
AMRMTokenIdentifier amrmTokenIdentifier = AuthorizeRequest();
ApplicationAttemptId appAttemptId = amrmTokenIdentifier.GetApplicationAttemptId();
ApplicationId applicationId = appAttemptId.GetApplicationId();
this.amLivelinessMonitor.ReceivedPing(appAttemptId);
/* check if its in cache */
ApplicationMasterService.AllocateResponseLock Lock = responseMap[appAttemptId];
if (Lock == null)
{
string message = "Application attempt " + appAttemptId + " doesn't exist in ApplicationMasterService cache.";
Log.Error(message);
throw new ApplicationAttemptNotFoundException(message);
}
lock (Lock)
{
AllocateResponse lastResponse = Lock.GetAllocateResponse();
if (!HasApplicationMasterRegistered(appAttemptId))
{
string message = "AM is not registered for known application attempt: " + appAttemptId
+ " or RM had restarted after AM registered . AM should re-register.";
Log.Info(message);
RMAuditLogger.LogFailure(this.rmContext.GetRMApps()[appAttemptId.GetApplicationId
()].GetUser(), RMAuditLogger.AuditConstants.AmAllocate, string.Empty, "ApplicationMasterService"
, message, applicationId, appAttemptId);
throw new ApplicationMasterNotRegisteredException(message);
}
if ((request.GetResponseId() + 1) == lastResponse.GetResponseId())
{
/* old heartbeat */
return(lastResponse);
}
else
{
if (request.GetResponseId() + 1 < lastResponse.GetResponseId())
{
string message = "Invalid responseId in AllocateRequest from application attempt: "
+ appAttemptId + ", expect responseId to be " + (lastResponse.GetResponseId() +
1);
throw new InvalidApplicationMasterRequestException(message);
}
}
//filter illegal progress values
float filteredProgress = request.GetProgress();
if (float.IsNaN(filteredProgress) || filteredProgress == float.NegativeInfinity ||
filteredProgress < 0)
{
request.SetProgress(0);
}
else
{
if (filteredProgress > 1 || filteredProgress == float.PositiveInfinity)
{
request.SetProgress(1);
}
}
// Send the status update to the appAttempt.
this.rmContext.GetDispatcher().GetEventHandler().Handle(new RMAppAttemptStatusupdateEvent
(appAttemptId, request.GetProgress()));
IList <ResourceRequest> ask = request.GetAskList();
IList <ContainerId> release = request.GetReleaseList();
ResourceBlacklistRequest blacklistRequest = request.GetResourceBlacklistRequest();
IList <string> blacklistAdditions = (blacklistRequest != null) ? blacklistRequest.
GetBlacklistAdditions() : Sharpen.Collections.EmptyList;
IList <string> blacklistRemovals = (blacklistRequest != null) ? blacklistRequest.GetBlacklistRemovals
() : Sharpen.Collections.EmptyList;
RMApp app = this.rmContext.GetRMApps()[applicationId];
// set label expression for Resource Requests if resourceName=ANY
ApplicationSubmissionContext asc = app.GetApplicationSubmissionContext();
foreach (ResourceRequest req in ask)
{
if (null == req.GetNodeLabelExpression() && ResourceRequest.Any.Equals(req.GetResourceName
()))
{
req.SetNodeLabelExpression(asc.GetNodeLabelExpression());
}
}
// sanity check
try
{
RMServerUtils.NormalizeAndValidateRequests(ask, rScheduler.GetMaximumResourceCapability
(), app.GetQueue(), rScheduler, rmContext);
}
catch (InvalidResourceRequestException e)
{
Log.Warn("Invalid resource ask by application " + appAttemptId, e);
throw;
}
try
{
RMServerUtils.ValidateBlacklistRequest(blacklistRequest);
}
catch (InvalidResourceBlacklistRequestException e)
{
Log.Warn("Invalid blacklist request by application " + appAttemptId, e);
throw;
}
// In the case of work-preserving AM restart, it's possible for the
// AM to release containers from the earlier attempt.
if (!app.GetApplicationSubmissionContext().GetKeepContainersAcrossApplicationAttempts
())
{
try
{
RMServerUtils.ValidateContainerReleaseRequest(release, appAttemptId);
}
catch (InvalidContainerReleaseException e)
{
Log.Warn("Invalid container release by application " + appAttemptId, e);
throw;
}
}
// Send new requests to appAttempt.
Allocation allocation = this.rScheduler.Allocate(appAttemptId, ask, release, blacklistAdditions
, blacklistRemovals);
if (!blacklistAdditions.IsEmpty() || !blacklistRemovals.IsEmpty())
{
Log.Info("blacklist are updated in Scheduler." + "blacklistAdditions: " + blacklistAdditions
+ ", " + "blacklistRemovals: " + blacklistRemovals);
}
RMAppAttempt appAttempt = app.GetRMAppAttempt(appAttemptId);
AllocateResponse allocateResponse = recordFactory.NewRecordInstance <AllocateResponse
>();
if (!allocation.GetContainers().IsEmpty())
{
allocateResponse.SetNMTokens(allocation.GetNMTokens());
}
// update the response with the deltas of node status changes
IList <RMNode> updatedNodes = new AList <RMNode>();
if (app.PullRMNodeUpdates(updatedNodes) > 0)
{
IList <NodeReport> updatedNodeReports = new AList <NodeReport>();
foreach (RMNode rmNode in updatedNodes)
{
SchedulerNodeReport schedulerNodeReport = rScheduler.GetNodeReport(rmNode.GetNodeID
());
Resource used = BuilderUtils.NewResource(0, 0);
int numContainers = 0;
if (schedulerNodeReport != null)
{
used = schedulerNodeReport.GetUsedResource();
numContainers = schedulerNodeReport.GetNumContainers();
}
NodeId nodeId = rmNode.GetNodeID();
NodeReport report = BuilderUtils.NewNodeReport(nodeId, rmNode.GetState(), rmNode.
GetHttpAddress(), rmNode.GetRackName(), used, rmNode.GetTotalCapability(), numContainers
, rmNode.GetHealthReport(), rmNode.GetLastHealthReportTime(), rmNode.GetNodeLabels
());
updatedNodeReports.AddItem(report);
}
allocateResponse.SetUpdatedNodes(updatedNodeReports);
}
allocateResponse.SetAllocatedContainers(allocation.GetContainers());
allocateResponse.SetCompletedContainersStatuses(appAttempt.PullJustFinishedContainers
());
allocateResponse.SetResponseId(lastResponse.GetResponseId() + 1);
allocateResponse.SetAvailableResources(allocation.GetResourceLimit());
allocateResponse.SetNumClusterNodes(this.rScheduler.GetNumClusterNodes());
// add preemption to the allocateResponse message (if any)
allocateResponse.SetPreemptionMessage(GeneratePreemptionMessage(allocation));
// update AMRMToken if the token is rolled-up
MasterKeyData nextMasterKey = this.rmContext.GetAMRMTokenSecretManager().GetNextMasterKeyData
();
if (nextMasterKey != null && nextMasterKey.GetMasterKey().GetKeyId() != amrmTokenIdentifier
.GetKeyId())
{
RMAppAttemptImpl appAttemptImpl = (RMAppAttemptImpl)appAttempt;
Org.Apache.Hadoop.Security.Token.Token <AMRMTokenIdentifier> amrmToken = appAttempt
.GetAMRMToken();
if (nextMasterKey.GetMasterKey().GetKeyId() != appAttemptImpl.GetAMRMTokenKeyId())
{
Log.Info("The AMRMToken has been rolled-over. Send new AMRMToken back" + " to application: "
+ applicationId);
amrmToken = rmContext.GetAMRMTokenSecretManager().CreateAndGetAMRMToken(appAttemptId
);
appAttemptImpl.SetAMRMToken(amrmToken);
}
allocateResponse.SetAMRMToken(Org.Apache.Hadoop.Yarn.Api.Records.Token.NewInstance
(amrmToken.GetIdentifier(), amrmToken.GetKind().ToString(), amrmToken.GetPassword
(), amrmToken.GetService().ToString()));
}
/*
* As we are updating the response inside the lock object so we don't
* need to worry about unregister call occurring in between (which
* removes the lock object).
*/
Lock.SetAllocateResponse(allocateResponse);
return(allocateResponse);
}
}