[
https://issues.apache.org/jira/browse/MESOS-1739?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15214389#comment-15214389
]
Greg Mann commented on MESOS-1739:
----------------------------------
For a first step, one approach would be to limit the scope of this work that
it's only possible to change the slave resources/attributes to a superset of
their previous values. As you see in review #25525, that was the original
approach because it makes things much simpler. In fact, though it may not be
*strictly* necessary to kill tasks if attributes are removed from a slave,
consider the following scenario: a task was started on a slave with a
particular attribute because that attribute indicates that the slave has access
to a certain region of the network. If the attribute is removed because this
region is no longer accessible, then it would make sense to kill the task which
has that dependency. In this case, the master could notify the framework that
the attributes were changed and the framework could take action as appropriate.
So I think it might make sense to initially just implement a reconfiguration to
a superset of previous attributes/resources. Do you have a specific use case
for changing attributes/resources to a *subset* of their previous values?
> Allow slave reconfiguration on restart
> --------------------------------------
>
> Key: MESOS-1739
> URL: https://issues.apache.org/jira/browse/MESOS-1739
> Project: Mesos
> Issue Type: Epic
> Reporter: Patrick Reilly
> Labels: external-volumes, mesosphere, myriad
>
> Make it so that either via a slave restart or a out of process "reconfigure"
> ping, the attributes and resources of a slave can be updated to be a superset
> of what they used to be.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)