You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@slider.apache.org by "ASF subversion and git services (JIRA)" <ji...@apache.org> on 2014/09/02 22:03:20 UTC
[jira] [Commented] (SLIDER-327) AM to move to optimized, queued
Node review and refresh action
[ https://issues.apache.org/jira/browse/SLIDER-327?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14118637#comment-14118637 ]
ASF subversion and git services commented on SLIDER-327:
--------------------------------------------------------
Commit 88b665db938325bf2c221614f7ffadefe71be632 in incubator-slider's branch refs/heads/develop from [~stevel@apache.org]
[ https://git-wip-us.apache.org/repos/asf?p=incubator-slider.git;h=88b665d ]
SLIDER-327 AM to move to optimized, queued Node review and refresh action
> AM to move to optimized, queued Node review and refresh action
> --------------------------------------------------------------
>
> Key: SLIDER-327
> URL: https://issues.apache.org/jira/browse/SLIDER-327
> Project: Slider
> Issue Type: Improvement
> Affects Versions: Slider 0.50
> Reporter: Steve Loughran
> Assignee: Steve Loughran
> Fix For: Slider 0.60
>
>
> The {{reviewRequestAndReleaseNodes()}} operation should be converted to an async action, with all operations/triggers for review posting it in with a short (1s? delay).
> The executor should not only call the AM, it can be clever about optimising its workload
> # if there are events in the queue marked as changing cluster size, the review should come after them.
> # if there is an existing review operations in the queue, there's no need to schedule a new one.
> Having a delay means that a sequence of container failure events coming in one after the other don't trigger lots of reviews
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)