[
https://issues.apache.org/jira/browse/YARN-6599?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16314907#comment-16314907
]
Wangda Tan commented on YARN-6599:
----------------------------------
[~kkaranasos], thanks for comments.
bq. I think we should not introduce node partitions in this JIRA. Not only for
the size of the patch, but because semantically it belongs to a different JIRA.
I can move it to a separate JIRA, but we need to support node partition in
first released version.
bq. What we did in YARN-7613 is to restrict the scope of a constraint to the
application that is defining it for the first version, so that we do not have
to deal with prefixes
This is not correct to me, different implementations can choose to support
*subset* of the SchedulingRequest, but we should not change semantics of the
API unless discussed. So a client should expect same result (not have to be
exactly same placement, but the constraint should be strictly respected) of a
SinglePlacementConstraint handled by PlacementProcessor or logic from this
patch. We should revisit changes done by YARN-7613 and follow what we have
decided in design doc.
bq. I can be convinced to remove the SELF target
Filed YARN-7709 for this.
> Support rich placement constraints in scheduler
> -----------------------------------------------
>
> Key: YARN-6599
> URL: https://issues.apache.org/jira/browse/YARN-6599
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Wangda Tan
> Assignee: Wangda Tan
> Attachments: YARN-6599-YARN-6592.003.patch,
> YARN-6599-YARN-6592.004.patch, YARN-6599-YARN-6592.005.patch,
> YARN-6599-YARN-6592.006.patch, YARN-6599-YARN-6592.007.patch,
> YARN-6599-YARN-6592.008.patch, YARN-6599-YARN-6592.wip.002.patch,
> YARN-6599.poc.001.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
