[jira] [Commented] (SLING-4481) Reduce the number of controller threads for queue

["Carsten Ziegeler (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SLING-4481?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14363412#comment-14363412 ] 

Carsten Ziegeler commented on SLING-4481:
-----------------------------------------

i've commited a first working implementation which at least passes all integration tests.
Instead of having a thread per queue, jobs are now started in the main queue manager thread when a new job is detected. This dramatically simplifies the queue handling.
Each queue starts as many jobs as it is allowed to (max thread handling). Whenever a job is finished, a new job is started. A period maintenance task tries to start job as well, just in case something got stuck somewhere.
The implementation needs some more work, especially the retry handling for ordered queues is not optimal yet.


> Reduce the number of controller threads for queue
> -------------------------------------------------
>
>                 Key: SLING-4481
>                 URL: https://issues.apache.org/jira/browse/SLING-4481
>             Project: Sling
>          Issue Type: Improvement
>          Components: Extensions
>    Affects Versions: Event 3.5.2
>            Reporter: Carsten Ziegeler
>            Assignee: Carsten Ziegeler
>
> Right now each queue is using a controller thread which is not taken from a pool. WIth a lot of queues this results in a lot of threads - although a queue is only started if there is a job for that queue, this still can be a significant number.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)