[
https://issues.apache.org/jira/browse/MESOS-1773?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14127214#comment-14127214
]
Niklas Quarfot Nielsen commented on MESOS-1773:
-----------------------------------------------
My suggestion would be to add it to the command info protobuf instead. A task
may not be a unix process and the field becomes hard to honor.
Another reason for tying it to the command info, is that we can reuse the logic
for everywhere where we shell out (custom health checks, using the command
executor, launching executor processes and so on).
The max({Task(1).Timeout, ..., Task(N).Timeout}) is hard to reason about and
you don't see all tasks (or timeouts up front) which means that you have to
bump it every time you see a new task (with higher timeout).
Thoughts?
> Make shutdown grace period configurable per task
> ------------------------------------------------
>
> Key: MESOS-1773
> URL: https://issues.apache.org/jira/browse/MESOS-1773
> Project: Mesos
> Issue Type: Improvement
> Components: general, slave
> Reporter: Alexander Rukletsov
> Labels: patch
>
> With [Issue 1571|https://issues.apache.org/jira/browse/MESOS-1571] fixed,
> shutdown grace periods on all levels are dependent on the slave flag. The
> next step is to make it configurable per task.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
