You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@sling.apache.org by "Stefan Egli (JIRA)" <ji...@apache.org> on 2014/02/14 17:56:24 UTC

[jira] [Resolved] (SLING-3389) [discovery] reduce write activity of discovery heartbeats

     [ https://issues.apache.org/jira/browse/SLING-3389?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Stefan Egli resolved SLING-3389.
--------------------------------

       Resolution: Fixed
    Fix Version/s: Discovery Impl 1.0.4

fixed. now announcement are considered valid, as long as the instance that holds it (parent in the repository) is alive, ie is in the clusterview. Reduces write activity of announcements to zero if no changes happen in topology.

> [discovery] reduce write activity of discovery heartbeats
> ---------------------------------------------------------
>
>                 Key: SLING-3389
>                 URL: https://issues.apache.org/jira/browse/SLING-3389
>             Project: Sling
>          Issue Type: Improvement
>          Components: Extensions
>    Affects Versions: Discovery Impl 1.0.2
>            Reporter: Stefan Egli
>            Assignee: Stefan Egli
>             Fix For: Discovery Impl 1.0.4
>
>
> As discussed on the list at [0] there are ways to reduce the #writes to the repository for topology connector announcements. This ticket is about 1) and 2) of mentioned thread. Here again for completeness sake:
>  1) store the announcement's lastHeartbeat value (ie created value) as a separate property instead of as part of the json. Plus only update the json if anything has changed.
>  2) don't store the announcement's lastHeartbeat (or created) at all - but instead change the logic such that an announcement is valid as long the container-instance (the instance which has received and stored the announcement) is alive. This increases reaction time slightly (in the worst case x2..) in case the containing instance and some instance which is part of the announcement crash within the same heartbeat interval.
> [0] - http://markmail.org/thread/2ev5sy3b3mr5klc5



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)