[jira] [Commented] (MESOS-1720) Slave should send exited executor message when the executor is never launched.

[Greg Mann (JIRA)]

    [ 
https://issues.apache.org/jira/browse/MESOS-1720?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16363865#comment-16363865
 ] 

Greg Mann commented on MESOS-1720:
----------------------------------

Patches on 1.5.x:
{code}
commit 2bdf4935b7929d0dce614d76461cddb991df89da
Author: Meng Zhu <m...@mesosphere.io>
Date:   Tue Feb 13 22:45:07 2018 -0800

    Fixed a bug where executor info lingers on master if failed to launch.

    Master relies on `ExitedExecutorMessage` from the agent to remove
    executor entries. However, this message won't be sent if an executor
    never actually launched (due to transient error), leaving executor
    info on the master and the executor's resources claimed.
    See MESOS-1720.

    This patch fixes this issue by sending the `ExitedExecutorMessage`
    from the agent if the executor is never launched.

    Review: https://reviews.apache.org/r/65449/
{code}
{code}
commit fb0e2f1f81b2256a76cae83893e2a69fdd91fcd7
Author: Meng Zhu <m...@mesosphere.io>
Date:   Tue Feb 13 22:45:03 2018 -0800

    Added helper function for the agent to send `ExitedExecutorMessage`.

    Review: https://reviews.apache.org/r/65446/
{code}
{code}
commit 10aa875df8947f8cbfb318820101984d99259070
Author: Meng Zhu <m...@mesosphere.io>
Date:   Tue Feb 13 22:44:58 2018 -0800

    Made master set `launch_executor` in the RunTask(Group)Message.

    By setting a new field `launch_executor` in the RunTask(Group)Message,
    the master is able to control executor creation on the agent.

    Also refactored the `addTask()` logic. Added two new functions:
    `isTaskLaunchExecutor()` checks if a task needs to launch an executor;
    `addExecutor()` adds an executor to the framework and slave.

    Review: https://reviews.apache.org/r/65504/
{code}
{code}
commit 08e0ceb84e4bf353e1f938482bb6766bf73310c7
Author: Meng Zhu <m...@mesosphere.io>
Date:   Tue Feb 13 22:44:48 2018 -0800

    Added new protobuf field `launch_executor` in RunTask(Group)Message.

    This boolean flag is used for the master to specify whether a
    new executor should be launched for the task or task group (with
    the exception of the command executor). This allows the master
    to control executor creation on the agent.

    Also updated the relevant message handlers and mock functions.

    Review: https://reviews.apache.org/r/65445/
{code}

> Slave should send exited executor message when the executor is never launched.
> ------------------------------------------------------------------------------
>
>                 Key: MESOS-1720
>                 URL: https://issues.apache.org/jira/browse/MESOS-1720
>             Project: Mesos
>          Issue Type: Bug
>          Components: agent, master
>            Reporter: Benjamin Mahler
>            Assignee: Meng Zhu
>            Priority: Major
>              Labels: mesosphere
>             Fix For: 1.6.0, 1.5.1
>
>
> When the slave sends TASK_LOST before launching an executor for a task, the 
> slave does not send an exited executor message to the master.
> Since the master receives no exited executor message, it still thinks the 
> executor's resources are consumed on the slave.
> One possible fix for this would be to send the exited executor message to the 
> master in these cases.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)