[
https://issues.apache.org/jira/browse/YARN-2312?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14143699#comment-14143699
]
Jason Lowe commented on YARN-2312:
----------------------------------
bq. One idea is to add id for upper 32 bits of container Id to ID class.
The ID class is used by much more than just JvmID objects. I'm not a fan of
making all IDs pay for this extra storage when we only need it for this one
case. I'd rather store the extra bits in JvmID.
Actually I don't think it's critical that JvmID derives from ID. We could have
JvmID store the long itself rather than try to hack an extra 4-bytes onto ID
and then need to explain why JvmID.getId doesn't do what one would expect.
> Marking ContainerId#getId as deprecated
> ---------------------------------------
>
> Key: YARN-2312
> URL: https://issues.apache.org/jira/browse/YARN-2312
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: resourcemanager
> Reporter: Tsuyoshi OZAWA
> Assignee: Tsuyoshi OZAWA
> Attachments: YARN-2312-wip.patch
>
>
> {{ContainerId#getId}} will only return partial value of containerId, only
> sequence number of container id without epoch, after YARN-2229. We should
> mark {{ContainerId#getId}} as deprecated and use
> {{ContainerId#getContainerId}} instead.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
