org.apache.hadoop.yarn.client.api.async

Class AMRMClientAsync<T extends org.apache.hadoop.yarn.client.api.AMRMClient.ContainerRequest>

java.lang.Object

org.apache.hadoop.service.AbstractService

org.apache.hadoop.yarn.client.api.async.AMRMClientAsync<T>

All Implemented Interfaces:

Closeable, AutoCloseable, Service

@InterfaceAudience.Public
@InterfaceStability.Stable
public abstract class AMRMClientAsync<T extends org.apache.hadoop.yarn.client.api.AMRMClient.ContainerRequest>
extends AbstractService

AMRMClientAsync handles communication with the ResourceManager and provides asynchronous updates on events such as container allocations and completions. It contains a thread that sends periodic heartbeats to the ResourceManager. It should be used by implementing a CallbackHandler:

 class MyCallbackHandler implements AMRMClientAsync.CallbackHandler {
   public void onContainersAllocated(List<Container> containers) {
     [run tasks on the containers]
   }
   
   public void onContainersCompleted(List<ContainerStatus> statuses) {
     [update progress, check whether app is done]
   }
   
   public void onNodesUpdated(List<NodeReport> updated) {}
   
   public void onReboot() {}
 }
 
 

The client's lifecycle should be managed similarly to the following:

 AMRMClientAsync asyncClient = 
     createAMRMClientAsync(appAttId, 1000, new MyCallbackhandler());
 asyncClient.init(conf);
 asyncClient.start();
 RegisterApplicationMasterResponse response = asyncClient
    .registerApplicationMaster(appMasterHostname, appMasterRpcPort,
       appMasterTrackingUrl);
 asyncClient.addContainerRequest(containerRequest);
 [... wait for application to complete]
 asyncClient.unregisterApplicationMaster(status, appMsg, trackingUrl);
 asyncClient.stop();
 
 

Field Summary

Fields

Modifier and Type	Field and Description
protected AMRMClient<T>	client
protected org.apache.hadoop.yarn.client.api.async.AMRMClientAsync.CallbackHandler	handler
protected AtomicInteger	heartbeatIntervalMs

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	AMRMClientAsync(AMRMClient<T> client, int intervalMs, org.apache.hadoop.yarn.client.api.async.AMRMClientAsync.CallbackHandler callbackHandler)
protected	AMRMClientAsync(int intervalMs, org.apache.hadoop.yarn.client.api.async.AMRMClientAsync.CallbackHandler callbackHandler)

Method Summary

Methods

Modifier and Type	Method and Description
abstract void	addContainerRequest(T req) Request containers for resources before calling allocate
static <T extends org.apache.hadoop.yarn.client.api.AMRMClient.ContainerRequest> AMRMClientAsync<T>	createAMRMClientAsync(AMRMClient<T> client, int intervalMs, org.apache.hadoop.yarn.client.api.async.AMRMClientAsync.CallbackHandler callbackHandler)
static <T extends org.apache.hadoop.yarn.client.api.AMRMClient.ContainerRequest> AMRMClientAsync<T>	createAMRMClientAsync(int intervalMs, org.apache.hadoop.yarn.client.api.async.AMRMClientAsync.CallbackHandler callbackHandler)
abstract Resource	getAvailableResources() Get the currently available resources in the cluster.
abstract int	getClusterNodeCount() Get the current number of nodes in the cluster.
abstract List<? extends Collection<T>>	getMatchingRequests(Priority priority, String resourceName, Resource capability)
abstract RegisterApplicationMasterResponse	registerApplicationMaster(String appHostName, int appHostPort, String appTrackingUrl) Registers this application master with the resource manager.
abstract void	releaseAssignedContainer(ContainerId containerId) Release containers assigned by the Resource Manager.
abstract void	removeContainerRequest(T req) Remove previous container request.
void	setHeartbeatInterval(int interval)
abstract void	unregisterApplicationMaster(FinalApplicationStatus appStatus, String appMessage, String appTrackingUrl) Unregister the application master.
void	waitFor(com.google.common.base.Supplier<Boolean> check) Wait for check to return true for each 1000 ms.
void	waitFor(com.google.common.base.Supplier<Boolean> check, int checkEveryMillis) Wait for check to return true for each checkEveryMillis ms.
void	waitFor(com.google.common.base.Supplier<Boolean> check, int checkEveryMillis, int logInterval) Wait for check to return true for each checkEveryMillis ms.

Methods inherited from class org.apache.hadoop.service.AbstractService

close, getBlockers, getConfig, getFailureCause, getFailureState, getLifecycleHistory, getName, getServiceState, getStartTime, init, isInState, noteFailure, putBlocker, registerGlobalListener, registerServiceListener, removeBlocker, serviceInit, serviceStart, serviceStop, setConfig, start, stop, toString, unregisterGlobalListener, unregisterServiceListener, waitForServiceToStop

Methods inherited from class java.lang.Object

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

Field Detail

client

protected final AMRMClient<T extends org.apache.hadoop.yarn.client.api.AMRMClient.ContainerRequest> client

handler

protected final org.apache.hadoop.yarn.client.api.async.AMRMClientAsync.CallbackHandler handler

heartbeatIntervalMs

protected final AtomicInteger heartbeatIntervalMs

Constructor Detail

AMRMClientAsync

protected AMRMClientAsync(int intervalMs,
               org.apache.hadoop.yarn.client.api.async.AMRMClientAsync.CallbackHandler callbackHandler)

AMRMClientAsync

@InterfaceAudience.Private
protected AMRMClientAsync(AMRMClient<T> client,
                                         int intervalMs,
                                         org.apache.hadoop.yarn.client.api.async.AMRMClientAsync.CallbackHandler callbackHandler)

Method Detail

createAMRMClientAsync

public static <T extends org.apache.hadoop.yarn.client.api.AMRMClient.ContainerRequest> AMRMClientAsync<T> createAMRMClientAsync(int intervalMs,
                                                                                                                 org.apache.hadoop.yarn.client.api.async.AMRMClientAsync.CallbackHandler callbackHandler)

createAMRMClientAsync

public static <T extends org.apache.hadoop.yarn.client.api.AMRMClient.ContainerRequest> AMRMClientAsync<T> createAMRMClientAsync(AMRMClient<T> client,
                                                                                                                 int intervalMs,
                                                                                                                 org.apache.hadoop.yarn.client.api.async.AMRMClientAsync.CallbackHandler callbackHandler)

setHeartbeatInterval

public void setHeartbeatInterval(int interval)

getMatchingRequests

public abstract List<? extends Collection<T>> getMatchingRequests(Priority priority,
                                                String resourceName,
                                                Resource capability)

registerApplicationMaster

public abstract RegisterApplicationMasterResponse registerApplicationMaster(String appHostName,
                                                          int appHostPort,
                                                          String appTrackingUrl)
                                                                     throws YarnException,
                                                                            IOException

Registers this application master with the resource manager. On successful registration, starts the heartbeating thread.

Throws:

YarnException

IOException

unregisterApplicationMaster

public abstract void unregisterApplicationMaster(FinalApplicationStatus appStatus,
                               String appMessage,
                               String appTrackingUrl)
                                          throws YarnException,
                                                 IOException

Unregister the application master. This must be called in the end.

Parameters:

appStatus - Success/Failure status of the master

appMessage - Diagnostics message on failure

appTrackingUrl - New URL to get master info

Throws:

YarnException

IOException

addContainerRequest

public abstract void addContainerRequest(T req)

Request containers for resources before calling allocate

Parameters:

req - Resource request

removeContainerRequest

public abstract void removeContainerRequest(T req)

Remove previous container request. The previous container request may have already been sent to the ResourceManager. So even after the remove request the app must be prepared to receive an allocation for the previous request even after the remove request

Parameters:

req - Resource request

releaseAssignedContainer

public abstract void releaseAssignedContainer(ContainerId containerId)

Release containers assigned by the Resource Manager. If the app cannot use the container or wants to give up the container then it can release them. The app needs to make new requests for the released resource capability if it still needs it. eg. it released non-local resources

Parameters:

containerId -

getAvailableResources

public abstract Resource getAvailableResources()

Get the currently available resources in the cluster. A valid value is available after a call to allocate has been made

Returns:

Currently available resources

getClusterNodeCount

public abstract int getClusterNodeCount()

Get the current number of nodes in the cluster. A valid values is available after a call to allocate has been made

Returns:

Current number of nodes in the cluster

waitFor

public void waitFor(com.google.common.base.Supplier<Boolean> check)
             throws InterruptedException

Wait for check to return true for each 1000 ms. See also waitFor(com.google.common.base.Supplier, int) and waitFor(com.google.common.base.Supplier, int, int)

Parameters:

check -

Throws:

InterruptedException

waitFor

public void waitFor(com.google.common.base.Supplier<Boolean> check,
           int checkEveryMillis)
             throws InterruptedException

Wait for check to return true for each checkEveryMillis ms. See also waitFor(com.google.common.base.Supplier, int, int)

Parameters:

check - user defined checker

checkEveryMillis - interval to call check

Throws:

InterruptedException

waitFor

public void waitFor(com.google.common.base.Supplier<Boolean> check,
           int checkEveryMillis,
           int logInterval)
             throws InterruptedException

Wait for check to return true for each checkEveryMillis ms. In the main loop, this method will log the message "waiting in main loop" for each logInterval times iteration to confirm the thread is alive.

Parameters:

check - user defined checker

checkEveryMillis - interval to call check

logInterval - interval to log for each

Throws:

InterruptedException