[
https://issues.apache.org/jira/browse/MESOS-2291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14343875#comment-14343875
]
Vinod Kone commented on MESOS-2291:
-----------------------------------
[~arojas] When this validation is done by the slave, it shouldn't abort but
rather shutdown the executor on receipt of TASK_STAGING (or any unknown state).
The CHECK() code in slave you referenced above is the state of the slave. It
has nothing to do with the state of the task status of an update.
Regarding the unacknowledged status and tasks, that is the most tricky part to
handle when we move away from the executor driver (towards http api). Right now
this information is leveraged by the executor to properly reconcile the state
with the slave upon a slave restart.
> Move executor driver validations to slave
> -----------------------------------------
>
> Key: MESOS-2291
> URL: https://issues.apache.org/jira/browse/MESOS-2291
> Project: Mesos
> Issue Type: Task
> Reporter: Vinod Kone
> Assignee: Alexander Rojas
>
> With HTTP API, the executor driver will no longer exist and hence all the
> validations should move to the slave.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
