[
https://issues.apache.org/jira/browse/YARN-3216?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14936488#comment-14936488
]
Wangda Tan commented on YARN-3216:
----------------------------------
Hi [~sunilg],
Thanks for working on this, not finished, some comments so far:
1) (minor) Could you make amlimit computation treat label=="" as a normal
label? Which could simplify logic. You can use a map to store computed
amlimit-by-partition to avoid duplicated computation.
2) (major) getAMResourceLimitPerPartition should uses partition.totalResource
(from RMNodeLabelsManager.getPartitionResource) instead of clusterResource.
3) (minor) ResourceUsage#getAllAMUsed is not used.
4) (major) LeafQueue#getNumActiveAppsPerPartition is a O(n) operation, should
be optimized, otherwise activateApplication becomes a O(n^2) operation.
> Max-AM-Resource-Percentage should respect node labels
> -----------------------------------------------------
>
> Key: YARN-3216
> URL: https://issues.apache.org/jira/browse/YARN-3216
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: resourcemanager
> Reporter: Wangda Tan
> Assignee: Sunil G
> Priority: Critical
> Attachments: 0001-YARN-3216.patch, 0002-YARN-3216.patch
>
>
> Currently, max-am-resource-percentage considers default_partition only. When
> a queue can access multiple partitions, we should be able to compute
> max-am-resource-percentage based on that.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
