edu.umass.cs.reconfiguration.reconfigurationprotocoltasks

Class CommitWorker<NodeIDType>

java.lang.Object

edu.umass.cs.reconfiguration.reconfigurationprotocoltasks.CommitWorker<NodeIDType>

Type Parameters:

NodeIDType - This class is not a protocol task, just a usual runnable worker. It is initiated at a reconfigurator in order to commit the coordinated RCRecordRequest packets. We need a task for this because simply invoking handleIncoming (that in turn calls paxos propose) does not suffice to ensure that the command will be eventually committed.

All Implemented Interfaces:

java.lang.Runnable

public class CommitWorker<NodeIDType>
extends java.lang.Object
implements java.lang.Runnable

Author:

V. Arun

Constructor Summary

Constructors

Constructor and Description
CommitWorker(AbstractReplicaCoordinator<?> coordinator, ReconfiguratorCallback callback)

Method Summary

Modifier and Type	Method and Description
void	close()
boolean	coordinate(RCRecordRequest<NodeIDType> request)
boolean	enqueueForExecution(RCRecordRequest<NodeIDType> request)
boolean	enqueueForExecution(RCRecordRequest<NodeIDType> request, long restartPeriod) Should be used sparingly only for non-default restart periods, not for all names.
boolean	executedCallback(RCRecordRequest<NodeIDType> request)
boolean	executedCallback(RCRecordRequest<NodeIDType> request, boolean handled) We process some failed (handled = false) execution notifications to solve the early notify issue with complete requests for service names, i.e., the notify arrives before the pending request is enqueued but the pending request has been obviated anyway.
long	getPeriod()
static void	main(java.lang.String[] args) Sanity checking equals and hashCode for RCRecordRequest.
void	run()
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Constructor Detail

CommitWorker

public CommitWorker(AbstractReplicaCoordinator<?> coordinator,
                    ReconfiguratorCallback callback)

Parameters:

coordinator -

callback -

Method Detail

run

public void run()

Specified by:

run in interface java.lang.Runnable

close

public void close()

enqueueForExecution

public boolean enqueueForExecution(RCRecordRequest<NodeIDType> request)

Parameters:

request -

Returns:

True if enqueued for coordination.

enqueueForExecution

public boolean enqueueForExecution(RCRecordRequest<NodeIDType> request,
                                   long restartPeriod)

Should be used sparingly only for non-default restart periods, not for all names.

Parameters:

request -

restartPeriod -

Returns:

True if enqueued for execution.

executedCallback

public boolean executedCallback(RCRecordRequest<NodeIDType> request)

Parameters:

request -

Returns:

True if a matching pending request was found.

executedCallback

public boolean executedCallback(RCRecordRequest<NodeIDType> request,
                                boolean handled)

We process some failed (handled = false) execution notifications to solve the early notify issue with complete requests for service names, i.e., the notify arrives before the pending request is enqueued but the pending request has been obviated anyway.

Parameters:

request -

handled -

Returns:

Refer executedCallback(RCRecordRequest).

coordinate

public boolean coordinate(RCRecordRequest<NodeIDType> request)

Parameters:

request -

Returns:

True if coordinated.

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

getPeriod

public long getPeriod()

Returns:

The re-attempt wait interval.

main

public static void main(java.lang.String[] args)

Sanity checking equals and hashCode for RCRecordRequest.

Parameters:

args -