Mailing-List: contact dev-help@activemq.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@activemq.apache.org
Date: Thu, 29 Aug 2013 08:03:52 +0000 (UTC)
From: "Andy Gumbrecht (JIRA)" <jira@apache.org>
To: dev@activemq.apache.org
Message-ID: <JIRA.12618706.1354609865410.52974.1377763432701@arcas>
In-Reply-To: <JIRA.12618706.1354609865410@arcas>
References: <JIRA.12618706.1354609865410@arcas>
Subject: [jira] [Reopened] (AMQ-4205) Expose thread pool configuration
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


     [ https://issues.apache.org/jira/browse/AMQ-4205?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Andy Gumbrecht reopened AMQ-4205:
---------------------------------


Adding a patch for a more useful 120 second idle timeout:

The use case is that if a pooled thread is not used within 10 seconds then it is destroyed. A new thread process is then created on a subsequent call.

There are three such pools configured with low idle times.

If polling of any kind occurs at 11 or more seconds (such as keep alive for example) then a new thread is created every time.

Watching processes (using VisualVM) of a relatively idle server one can see excessive thread creation and termination due to these low timeouts.

If the timeout is not going to be exposed for configuration then I would at least suggest a more reasonable 120 seconds.
                
> Expose thread pool configuration
> --------------------------------
>
>                 Key: AMQ-4205
>                 URL: https://issues.apache.org/jira/browse/AMQ-4205
>             Project: ActiveMQ
>          Issue Type: Improvement
>          Components: Broker, Transport
>    Affects Versions: 5.8.0
>         Environment: NA
>            Reporter: Andy Gumbrecht
>            Priority: Minor
>              Labels: Thread, pool
>             Fix For: 5.9.0
>
>   Original Estimate: 24h
>  Remaining Estimate: 24h
>
> Default thread pool timeout can be too low which in turn causes excessive thread death and re-creation in AbstractInactivityMonitor.java (default 10 seconds) and TaskRunnerFactory.java (default 30 seconds).
> It would be great if these pools were configurable.
> The target to aim for is that pooled threads never die under normal load ('normal' being use case specific). 
> See http://activemq.2283324.n4.nabble.com/InactivityMonitor-Creating-too-frequent-threads-td4656752.html#a4659878

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira