[
https://issues.apache.org/jira/browse/YARN-4138?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15133627#comment-15133627
]
sandflee commented on YARN-4138:
--------------------------------
Hi, [~mding], there may some cases not user/app error,
1, increase resource from 1G to 5G
2, get a 5G token and send it to NM
3, before container increased msg come to RM, AM decrease resource from 5G ->
4G, (last confirmed is 1G)
4, container increase msg come to RM, in current logic, we don't confirm it,
just send a decrease msg to NM
5, container increase timer expire, the resource is rolled back to 1G
in 4, maybe We could set lastConfirmed to 4G.
> Roll back container resource allocation after resource increase token expires
> -----------------------------------------------------------------------------
>
> Key: YARN-4138
> URL: https://issues.apache.org/jira/browse/YARN-4138
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: api, nodemanager, resourcemanager
> Reporter: MENG DING
> Assignee: MENG DING
> Attachments: YARN-4138-YARN-1197.1.patch,
> YARN-4138-YARN-1197.2.patch, YARN-4138.3.patch, YARN-4138.4.patch,
> YARN-4138.5.patch
>
>
> In YARN-1651, after container resource increase token expires, the running
> container is killed.
> This ticket will change the behavior such that when a container resource
> increase token expires, the resource allocation of the container will be
> reverted back to the value before the increase.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
