[
https://issues.apache.org/jira/browse/SLIDER-958?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15902022#comment-15902022
]
Gour Saha commented on SLIDER-958:
----------------------------------
After SLIDER-1184 has been fixed, app owners should set
*_yarn.container.failure.threshold_* to 0 during the upgrade command. The AM
will then not count the container kills during upgrade as failures.
> the application is failed by yarn.container.failure.threshold when upgrading
> the application
> --------------------------------------------------------------------------------------------
>
> Key: SLIDER-958
> URL: https://issues.apache.org/jira/browse/SLIDER-958
> Project: Slider
> Issue Type: Bug
> Affects Versions: Slider 0.80
> Reporter: kyungwan nam
> Assignee: Gour Saha
> Fix For: Slider 1.0.0
>
>
> There is a hbase-app that the number of the regionservers is 10.
> (yarn.component.instances=10)
> I upgraded each of the regionservers one by one using “slider upgrade
> —containers”.
> But the hbase-app was failed when I try to upgrade the 6th regionserver.
> the default of yarn.container.failure.threshold is 5. that’s because the
> hbase-app was failed.
> I think that a failure by upgrading should be excepted from the counting the
> failure of the role.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
