[
https://issues.apache.org/jira/browse/ARTEMIS-2240?focusedWorklogId=190835&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-190835
]
ASF GitHub Bot logged work on ARTEMIS-2240:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 28/Jan/19 08:42
Start Date: 28/Jan/19 08:42
Worklog Time Spent: 10m
Work Description: michaelandrepearce commented on issue #2524:
ARTEMIS-2240 ActiveMQThreadPoolExecutor should use LinkedTransferQueue
URL: https://github.com/apache/activemq-artemis/pull/2524#issuecomment-458039733
@franz1981 a slight improvement, i spotted old master wasn't honoring the
pre 2.4 original behaviour of on shutdown, rejection would then use default
policy of the executor (AbortPolicy) which would throw rejection exception. a
simple correction.
-->
```
/*
* ActiveMQThreadPoolExecutor: a special ThreadPoolExecutor that combines
* the benefits of a cached executor and a fixed size executor.
* Similar to a cached executor, threads exceeding the core size are only
created on demand,
* and will be removed after idling for a specified keep time.
* But in contrast to a standard cached executor, tasks are queued if the
* maximum pool size if reached, instead of rejected.
*/
public class ActiveMQThreadPoolExecutor extends ThreadPoolExecutor {
/**
* The default rejected execution handler
*/
private static final RejectedExecutionHandler defaultHandler = new
AbortPolicy();
// Handler executed when a task is submitted and a new thread cannot be
created (because maxSize was reached)
// It queues the task on the executors's queue (using the add() method,
see ThreadPoolQueue class below)
private static class QueueExecutionHandler implements
RejectedExecutionHandler {
private final RejectedExecutionHandler handler;
private QueueExecutionHandler(RejectedExecutionHandler handler) {
if (handler == null)
throw new NullPointerException();
this.handler = handler;
}
@Override
public void rejectedExecution(Runnable r, ThreadPoolExecutor executor)
{
if (executor.isShutdown() || !executor.getQueue().add(r)) {
handler.rejectedExecution(r, executor);
}
}
}
// A specialized LinkedBlockingQueue that takes new elements by calling
add() but not offer()
// This is to force the ThreadPoolExecutor to always create new threads
and never queue
private static class ThreadPoolQueue extends
LinkedTransferQueue<Runnable> {
@Override
public boolean offer(Runnable runnable) {
return tryTransfer(runnable);
}
@Override
public boolean add(Runnable runnable) {
return super.offer( runnable );
}
}
public ActiveMQThreadPoolExecutor(int coreSize, int maxSize, long keep,
TimeUnit keepUnits, ThreadFactory factory) {
this(coreSize, maxSize, keep, keepUnits, factory, defaultHandler);
}
public ActiveMQThreadPoolExecutor(int coreSize, int maxSize, long keep,
TimeUnit keepUnits, ThreadFactory factory, RejectedExecutionHandler handler) {
super(coreSize, maxSize, keep, keepUnits, new ThreadPoolQueue(),
factory, new QueueExecutionHandler(handler));
}
}
```
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 190835)
Time Spent: 2h 10m (was: 2h)
> ActiveMQThreadPoolExecutor should use LinkedTransferQueue
> ---------------------------------------------------------
>
> Key: ARTEMIS-2240
> URL: https://issues.apache.org/jira/browse/ARTEMIS-2240
> Project: ActiveMQ Artemis
> Issue Type: Improvement
> Reporter: Francesco Nigro
> Assignee: Francesco Nigro
> Priority: Major
> Time Spent: 2h 10m
> Remaining Estimate: 0h
>
> ActiveMQThreadPoolExecutor should use LinkedTransferQueue to reduce cost over
> contention, given that internally it isn't using ReentrantLock as
> LinkedBlockingQueue.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
