You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@mesos.apache.org by "Alexander Rukletsov (JIRA)" <ji...@apache.org> on 2014/09/01 12:16:20 UTC
[jira] [Commented] (MESOS-1736) Completed tasks shown as running
[ https://issues.apache.org/jira/browse/MESOS-1736?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14117277#comment-14117277 ]
Alexander Rukletsov commented on MESOS-1736:
--------------------------------------------
Yep, it looks like a task's status update in {{Master::removeFramework()}} would suffice. The patch is being tested.
However, stopping the framework launches the executor graceful shutdown mechanism, that, given [escalation is configurable |https://issues.apache.org/jira/browse/MESOS-1571], gives the opportunity for the tasks to finish. If we mark these tasks as killed and move to completed, we loose the chance to collect the possible finished status. I would propose either to initiate the hard shutdown for executors in this case, or wait for the actual task statuses from terminated or killed executors.
> Completed tasks shown as running
> --------------------------------
>
> Key: MESOS-1736
> URL: https://issues.apache.org/jira/browse/MESOS-1736
> Project: Mesos
> Issue Type: Bug
> Affects Versions: 0.21.0
> Reporter: Alexander Rukletsov
> Assignee: Alexander Rukletsov
> Priority: Minor
> Labels: newbie
>
> When a framework is shutdown by calling {{driver.stop()}} from the scheduler, running tasks appear as completed in mesos webui but with RUNNING state. {{tasks.json}} and {{state.json}} have the same issue as webui. Slave doesn't show any anomalies: [log|https://gist.github.com/rukletsov/09e8ddf2ee5d02ab06fe].
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)