[
https://issues.apache.org/jira/browse/DISPATCH-1281?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16786036#comment-16786036
]
ASF GitHub Bot commented on DISPATCH-1281:
------------------------------------------
ted-ross commented on pull request #457: DISPATCH-1281 - Batch the cleanup of
freed messages so they share a s…
URL: https://github.com/apache/qpid-dispatch/pull/457
…ingle unit of general-work per core-thread action sweep.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
> Performance - Batch the freeing of messages from the core thread
> ----------------------------------------------------------------
>
> Key: DISPATCH-1281
> URL: https://issues.apache.org/jira/browse/DISPATCH-1281
> Project: Qpid Dispatch
> Issue Type: Improvement
> Components: Router Node
> Reporter: Ted Ross
> Assignee: Ted Ross
> Priority: Major
> Fix For: 1.6.0
>
>
> Profiling has identified a disproportionate amount of core-thread time being
> spent in qdr_delete_delivery_internal_CT. This is due to the fact that every
> in-core delivery-free results in the posting of one unit of general work
> (work to be performed by an I/O thread without any association with a
> particular connection).
> This improvement batches the freed messages to be handled at one time, with
> one unit of general work, at the end of the processing of each set of
> core-thread actions.
> The result is a greater degree of batching of message-cleanups when the core
> thread is busier. This will free up core-thread CPU capacity for when the
> router is the busiest.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
