You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@sling.apache.org by "Carsten Ziegeler (JIRA)" <ji...@apache.org> on 2013/04/24 10:39:25 UTC

[jira] [Resolved] (SLING-2831) Use new discovery API

     [ https://issues.apache.org/jira/browse/SLING-2831?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Carsten Ziegeler resolved SLING-2831.
-------------------------------------

    Resolution: Fixed

All parts of the eventing now use the new discovery api. We handle crashed/stopped instances, changes in the capabilities of an instance and queue configuration changes.
After a change, there is a delay until the manager checks for changes. This keeps the changes to a minimum, e.g. if a bundle is just updated
                
> Use new discovery API
> ---------------------
>
>                 Key: SLING-2831
>                 URL: https://issues.apache.org/jira/browse/SLING-2831
>             Project: Sling
>          Issue Type: Improvement
>          Components: Extensions
>    Affects Versions: Extensions Event 3.1.4
>            Reporter: Carsten Ziegeler
>            Assignee: Carsten Ziegeler
>             Fix For: Extensions Event 3.2.0
>
>
> We should leverage the new discovery api in several ways:
> With SLING-2829 we have an API which allows to detect the capabilities of an instance (= the jobs the instance can process). The discovery API provides a way to distribute this information across the cluster. The job handling should use this information to distribute the jobs to instances which are capable of handling the jobs.
> I refer to this as the push based job distributed: once a job is created, the receiving instance can use the capabilities and push the job to an instance which is able to process this. This allows job processing without any additional overhead like locking etc.
> With the new topology view of the discovery api, we can handle crashing/stopped instances, changes in capabilities etc. As these events usually require redistribution, we can use the leader election of the topology view to let only one instance do the task

--
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