org.apache.hadoop.yarn.server.api.protocolrecords
Class RegisterNodeManagerRequest

java.lang.Object
  org.apache.hadoop.yarn.server.api.protocolrecords.RegisterNodeManagerRequest

Direct Known Subclasses:

RegisterNodeManagerRequestPBImpl

public abstract class RegisterNodeManagerRequest
extends Object

Constructor Summary

RegisterNodeManagerRequest()

Method Summary

abstract int	getHttpPort()
abstract List<NMContainerStatus>	getNMContainerStatuses()
abstract String	getNMVersion()
abstract org.apache.hadoop.yarn.api.records.NodeId	getNodeId()
abstract org.apache.hadoop.yarn.api.records.Resource	getResource()
abstract List<org.apache.hadoop.yarn.api.records.ApplicationId>	getRunningApplications() We introduce this here because currently YARN RM doesn't persist nodes info for application running.
static RegisterNodeManagerRequest	newInstance(org.apache.hadoop.yarn.api.records.NodeId nodeId, int httpPort, org.apache.hadoop.yarn.api.records.Resource resource, String nodeManagerVersionId, List<NMContainerStatus> containerStatuses, List<org.apache.hadoop.yarn.api.records.ApplicationId> runningApplications)
abstract void	setContainerStatuses(List<NMContainerStatus> containerStatuses)
abstract void	setHttpPort(int port)
abstract void	setNMVersion(String version)
abstract void	setNodeId(org.apache.hadoop.yarn.api.records.NodeId nodeId)
abstract void	setResource(org.apache.hadoop.yarn.api.records.Resource resource)
abstract void	setRunningApplications(List<org.apache.hadoop.yarn.api.records.ApplicationId> runningApplications) Setter for getRunningApplications()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

RegisterNodeManagerRequest

public RegisterNodeManagerRequest()

Method Detail

newInstance

public static RegisterNodeManagerRequest newInstance(org.apache.hadoop.yarn.api.records.NodeId nodeId,
                                                     int httpPort,
                                                     org.apache.hadoop.yarn.api.records.Resource resource,
                                                     String nodeManagerVersionId,
                                                     List<NMContainerStatus> containerStatuses,
                                                     List<org.apache.hadoop.yarn.api.records.ApplicationId> runningApplications)

getNodeId

public abstract org.apache.hadoop.yarn.api.records.NodeId getNodeId()

getHttpPort

public abstract int getHttpPort()

getResource

public abstract org.apache.hadoop.yarn.api.records.Resource getResource()

getNMVersion

public abstract String getNMVersion()

getNMContainerStatuses

public abstract List<NMContainerStatus> getNMContainerStatuses()

getRunningApplications

public abstract List<org.apache.hadoop.yarn.api.records.ApplicationId> getRunningApplications()

We introduce this here because currently YARN RM doesn't persist nodes info for application running. When RM restart happened, we cannot determinate if a node should do application cleanup (like log-aggregation, status update, etc.) or not.

When we have this running application list in node manager register request, we can recover nodes info for running applications. And then we can take actions accordingly

Returns:

running application list in this node

setNodeId

public abstract void setNodeId(org.apache.hadoop.yarn.api.records.NodeId nodeId)

setHttpPort

public abstract void setHttpPort(int port)

setResource

public abstract void setResource(org.apache.hadoop.yarn.api.records.Resource resource)

setNMVersion

public abstract void setNMVersion(String version)

setContainerStatuses

public abstract void setContainerStatuses(List<NMContainerStatus> containerStatuses)

setRunningApplications

public abstract void setRunningApplications(List<org.apache.hadoop.yarn.api.records.ApplicationId> runningApplications)

Setter for getRunningApplications()

Parameters:

runningApplications - running application in this node