[
https://issues.apache.org/jira/browse/MESOS-621?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13732259#comment-13732259
]
Vinod Kone commented on MESOS-621:
----------------------------------
I see. Thanks for the explanation Thomas. That makes sense.
I still think it makes more sense to remove the framework allocations when the
slave is removed, rather than making allocator do the extra work of
allocations, sending the offer to the master and then recovering them.
Also, FWICT, Allocator::frameworkRemoved() does remove the allocations for the
framework right away. We only call Allocator::resourcesRecovered() in
Master::offer() when a framework is removed to handle a race.
For me, Allocator::frameworkAdded() and frameworkRemoved() look complementary,
whereas Allocator::slaveAdded() and Allocator::slaveRemoved() do not. For the
latter, unless there is a good reason, I think we should make them
complementary. It is easy to reason about.
Thoughts?
> HierarchicalAllocator::slaveRemoved doesn't properly handle framework
> allocations/resources
> -------------------------------------------------------------------------------------------
>
> Key: MESOS-621
> URL: https://issues.apache.org/jira/browse/MESOS-621
> Project: Mesos
> Issue Type: Bug
> Reporter: Vinod Kone
> Assignee: Thomas Marshall
> Fix For: 0.14.0
>
>
> Currently a slaveRemoved() simply removes the slave from 'slaves' map and
> slave's resources from 'roleSorter'. Looking at resourcesRecovered(), more
> things need to be done when a slave is removed (e.g., framework
> unallocations).
> It would be nice to fix this and have a test for this.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
