You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@stratos.apache.org by "Nirmal Fernando (JIRA)" <ji...@apache.org> on 2014/04/23 06:20:15 UTC

[jira] [Resolved] (STRATOS-625) Error in scale down logic

     [ https://issues.apache.org/jira/browse/STRATOS-625?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Nirmal Fernando resolved STRATOS-625.
-------------------------------------

    Resolution: Fixed

> Error in scale down logic 
> --------------------------
>
>                 Key: STRATOS-625
>                 URL: https://issues.apache.org/jira/browse/STRATOS-625
>             Project: Stratos
>          Issue Type: Bug
>          Components: Autoscaler
>    Affects Versions: 4.0.0 Alpha
>            Reporter: Nirmal Fernando
>            Assignee: Nirmal Fernando
>            Priority: Blocker
>             Fix For: 4.0.0 RC1
>
>
> Seems like following current scale down rule is not accurate.
> scaleDown : Boolean() from ((rifReset && (rifPredictedValue < rifAverageLimit * 0.1)) || (mcReset && (mcPredictedValue < mcAverageLimit * 0.1)) || (laReset && (laPredictedValue < laAverageLimit * 0.1)))
> Note the 'OR' conditions. If you read this, it'll mean that system would scale down if any of the factors resulted to drop down than its threshold, despite higher requests in one of the factors.
> IMO we should use 'AND's instead of 'OR's.
> Proposed fix:
> scaleDown : Boolean() from ((rifReset && (rifPredictedValue < rifAverageLimit * 0.1)) && (mcReset && (mcPredictedValue < mcAverageLimit * 0.1)) && (laReset && (laPredictedValue < laAverageLimit * 0.1)))



--
This message was sent by Atlassian JIRA
(v6.2#6252)