You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@slider.apache.org by "Steve Loughran (JIRA)" <ji...@apache.org> on 2015/05/02 22:25:06 UTC
[jira] [Commented] (SLIDER-856) slider needs to treat pre-emption
events as not-a-real-failure
[ https://issues.apache.org/jira/browse/SLIDER-856?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14525464#comment-14525464 ]
Steve Loughran commented on SLIDER-856:
---------------------------------------
Delving into the code, slider was treating all unexpected container exits as an app failure, incrementing failure counts of app & node.
The rework looks at the root cause, and treats the following as possible/definite node failures to count
* general app exit
* failure no disks
App failures are
* general app exit
* pmem or vmem limits exceeded
Preemption is now not counted by either; it's simply something that gets counted
> slider needs to treat pre-emption events as not-a-real-failure
> --------------------------------------------------------------
>
> Key: SLIDER-856
> URL: https://issues.apache.org/jira/browse/SLIDER-856
> Project: Slider
> Issue Type: Bug
> Components: appmaster
> Affects Versions: Slider 0.70
> Reporter: Steve Loughran
> Assignee: Steve Loughran
> Priority: Critical
> Fix For: Slider 0.80
>
>
> look at the exit codes and don't treat pre-emption as failures
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)