org.apache.flink.runtime.jobmanager

Class EventCollector

java.lang.Object

java.util.TimerTask

org.apache.flink.runtime.jobmanager.EventCollector

All Implemented Interfaces:

Runnable, ProfilingListener

public final class EventCollector
extends TimerTask
implements ProfilingListener

The event collector collects events which occurred during the execution of a job and prepares them for being fetched by a client. The collected events have an expiration time. In a configurable interval the event collector removes all intervals which are older than the interval.

Constructor Summary

Constructors

Constructor and Description
EventCollector(int clientQueryInterval) Constructs a new event collector and starts its background cleanup routine.

Method Summary

Methods

Modifier and Type	Method and Description
void	getEventsForJob(JobID jobID, List<AbstractEvent> eventList, boolean includeManagementEvents) Retrieves and adds the collected events for the job with the given job ID to the provided list.
ExecutionGraph	getManagementGraph(JobID jobID) Returns the execution graph object for the job with the given ID from the map of recently added graphs.
void	getRecentJobs(List<RecentJobEvent> eventList)
void	processProfilingEvents(ProfilingEvent profilingEvent)
void	registerArchivist(ArchiveListener al) Register Archivist to archive
void	registerJob(ExecutionGraph executionGraph, boolean profilingAvailable, long submissionTimestamp) Registers a job in form of its execution graph representation with the job progress collector.
void	run() This method will periodically be called to clean up expired collected events.
void	shutdown() Stops the timer thread and cleans up the data structure which stores the collected events.

Methods inherited from class java.util.TimerTask

cancel, scheduledExecutionTime

Methods inherited from class java.lang.Object

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

Constructor Detail

EventCollector

public EventCollector(int clientQueryInterval)

Constructs a new event collector and starts its background cleanup routine.

Parameters:

clientQueryInterval - the interval with which clients query for events

Method Detail

getEventsForJob

public void getEventsForJob(JobID jobID,
                   List<AbstractEvent> eventList,
                   boolean includeManagementEvents)

Retrieves and adds the collected events for the job with the given job ID to the provided list.

Parameters:

jobID - the ID of the job to retrieve the events for

eventList - the list to which the events shall be added

includeManagementEvents - true if ManagementEvent objects shall be added to the list as well, false otherwise

getRecentJobs

public void getRecentJobs(List<RecentJobEvent> eventList)

shutdown

public void shutdown()

Stops the timer thread and cleans up the data structure which stores the collected events.

registerJob

public void registerJob(ExecutionGraph executionGraph,
               boolean profilingAvailable,
               long submissionTimestamp)

Registers a job in form of its execution graph representation with the job progress collector. The collector will subscribe to state changes of the individual subtasks. A separate de-registration is not necessary since the job progress collector periodically discards outdated progress information.

Parameters:

executionGraph - the execution graph representing the job

profilingAvailable - indicates if profiling data is available for this job

submissionTimestamp - the submission time stamp of the job

run

public void run()

This method will periodically be called to clean up expired collected events.

Specified by:

run in interface Runnable

Specified by:

run in class TimerTask

processProfilingEvents

public void processProfilingEvents(ProfilingEvent profilingEvent)

Specified by:

processProfilingEvents in interface ProfilingListener

getManagementGraph

public ExecutionGraph getManagementGraph(JobID jobID)

Returns the execution graph object for the job with the given ID from the map of recently added graphs.

Parameters:

jobID - The ID of the job the management graph shall be retrieved for

Returns:

the management graph for the job with the given ID or null if no such graph exists

registerArchivist

public void registerArchivist(ArchiveListener al)

Register Archivist to archive