Steve Loughran created SLIDER-808:
-------------------------------------

             Summary: AM to not-overreact to pre-emption or NM failures
                 Key: SLIDER-808
                 URL: https://issues.apache.org/jira/browse/SLIDER-808
             Project: Slider
          Issue Type: Improvement
            Reporter: Steve Loughran


AM should look at the values of {{ContainerExitStatus}} to decide whether to 
blacklist nodes/increment failure count windows based on what actually happened.

* ABORTED => trouble
* PMEM and VMEM exceeded: trouble. 
* PREEMPTED and DISKS_FAILED => don't increment counts



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to