[
https://issues.apache.org/jira/browse/YARN-6144?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15854760#comment-15854760
]
Karthik Kambatla commented on YARN-6144:
----------------------------------------
Thanks for finding and working on this, Miklos.
Comments on the patch:
# The debug logging does not add much information. Maybe, we should skip it.
Sorry for vacillating on this.
# Test:
## Not all asserts are to identify a regression. That is okay, but I just
wanted to call that out.
## On each of the asserts, can we use assertEquals instead of assertTrue so
test failures become easier to triage.
## Can we add a message for each of the new asserts?
## And, can we use Resources.none() instead of creating a new instance every
time?
> FairScheduler: preempted resources can become negative
> ------------------------------------------------------
>
> Key: YARN-6144
> URL: https://issues.apache.org/jira/browse/YARN-6144
> Project: Hadoop YARN
> Issue Type: Bug
> Components: fairscheduler, resourcemanager
> Affects Versions: 3.0.0-alpha2
> Reporter: Miklos Szegedi
> Assignee: Miklos Szegedi
> Priority: Blocker
> Attachments: Screen Shot 2017-02-02 at 12.49.14 PM.png,
> YARN-6144.000.patch, YARN-6144.001.patch
>
>
> {{preemptContainers()}} calls {{trackContainerForPreemption()}} to collect
> the list of containers and resources that were preempted for an application.
> Later the list is reduced when {{containerCompleted()}} calls
> {{untrackContainerForPreemption()}}. The bug is that the resource variable
> {{preemptedResources}} is subtracted, not just when the container was
> preempted but also when it has completed successfully. This causes that we
> return an incorrect value in {{getResourceUsage()}}.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
